<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Jagruti Pawashe on Medium]]></title>
        <description><![CDATA[Stories by Jagruti Pawashe on Medium]]></description>
        <link>https://medium.com/@pawashejagruti?source=rss-73530dc3fd0f------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*YUjgsMFMjHYsZJRQ</url>
            <title>Stories by Jagruti Pawashe on Medium</title>
            <link>https://medium.com/@pawashejagruti?source=rss-73530dc3fd0f------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 23:37:41 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@pawashejagruti/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[From Words to Intelligence: The Rise of the Transformers]]></title>
            <link>https://pawashejagruti.medium.com/from-words-to-intelligence-the-rise-of-the-transformers-c1a2f9bcd841?source=rss-73530dc3fd0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/c1a2f9bcd841</guid>
            <category><![CDATA[transformers]]></category>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[lstm]]></category>
            <category><![CDATA[genai]]></category>
            <category><![CDATA[rnn]]></category>
            <dc:creator><![CDATA[Jagruti Pawashe]]></dc:creator>
            <pubDate>Tue, 23 Jun 2026 08:46:41 GMT</pubDate>
            <atom:updated>2026-06-23T10:57:51.464Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/801/1*7Jbg-m9UNKXw-Mw_o-9BJQ.gif" /></figure><h3>Introduction:</h3><p>Imagine asking a machine to write a poem, explain quantum physics, generate a piece of code, and even help you prepare for an interview — all within a few seconds.</p><p>A decade ago, this would have sounded like science fiction.</p><p>Yet today, millions of people interact daily with AI systems like ChatGPT, often forgetting that behind these seemingly intelligent conversations lies an architectural breakthrough that is less than a decade old: the <strong>Transformer</strong>.</p><p>Interestingly, the Transformer wasn’t built to create chatbots or write code. It was introduced to solve a fundamental problem in language understanding — how can a machine know which words matter the most in a sentence?</p><p>The answer to that question sparked a revolution that transformed Natural Language Processing and gave birth to the era of Generative AI.</p><h3>Why Transformers Changed AI Forever?</h3><blockquote>“The best innovations are often born from the limitations of existing technologies.”</blockquote><p>Today, AI can write essays, generate code, summarize research papers, and even hold conversations that feel remarkably human. But just a few years ago, machines struggled with something much simpler — “<strong>remembering context</strong>.”</p><p>The journey from Recurrent Neural Networks (RNNs) to Transformers is a story of overcoming one fundamental challenge:</p><blockquote>How can a machine understand relationships between words that are far apart?</blockquote><p><strong>The Era of Sequential Thinking:</strong></p><p>Before Transformers, the world of Natural Language Processing (NLP) was dominated by <strong>Recurrent Neural Networks (RNNs)</strong>. The idea behind RNNs was simple and intuitive. To understand a sentence, read it one word at a time.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*1U8H9EZiDqfylJU7Im23Ag.gif" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*q1G8OaLVqRLBjUvOQWfbOw.gif" /></figure><p>Consider the sentence:</p><blockquote><em>“The movie that I watched with my friends yesterday was amazing.”</em></blockquote><p>An RNN processes it like this:</p><pre>The → movie → that → I → watched → ... → amazing</pre><p>At every step, it carries information from the previous word to the next.</p><p>This sounds reasonable because that’s how humans read sentences, too.</p><p>Unfortunately, for machines, this approach came with a major problem.</p><p><strong>The Memory Problem:</strong></p><p>Imagine I tell you:</p><blockquote><em>“The animal didn’t cross the street because it was tired.”</em></blockquote><p>Who was tired? → The animal.</p><p>To answer this, the model needs to remember information from several words earlier. As sentences become longer, RNNs start forgetting important information. It’s like playing the game of “Telephone,” where a message gets distorted as it passes from one person to another.</p><p>By the time an RNN reaches the end of a long sentence, much of the earlier context has faded away. This phenomenon is known as the <strong>Long-Term Dependency Problem.</strong></p><p><strong>The Vanishing Gradient Problem:</strong></p><p>The issue became even more severe during training. RNNs learn through a process called <strong>Backpropagation Through Time (BPTT)</strong>.</p><p>As errors are propagated backward through many time steps, the gradients often become extremely small. This is known as the <strong>Vanishing Gradient Problem</strong>.</p><p>As a result:</p><ul><li>Earlier words stop influencing learning.</li><li>The network struggles to capture long-range relationships.</li><li>Training becomes unstable.</li></ul><p>The model essentially develops a very short memory.</p><p><strong>LSTMs: A Better Memory System</strong></p><p>To solve this, researchers introduced <strong>Long Short-Term Memory Networks (LSTMs). </strong>LSTMs came with a brilliant idea:</p><blockquote><em>“What if the model could decide what to remember and what to forget?”</em></blockquote><p>They introduced:</p><ul><li>Forget Gate</li><li>Input Gate</li><li>Output Gate</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*n-IgHZM5baBUjq0T7RYDBw.gif" /></figure><p>These gates acted like filters, controlling the flow of information.</p><p>LSTMs significantly improved performance on many NLP tasks.</p><p>For a while, it looked like the problem had been solved. But it hadn’t.</p><h4><strong>LSTMs Were Better… But Not Perfect</strong></h4><p>Although LSTMs could remember longer contexts, they still suffered from several limitations.</p><h3>1. Sequential Processing</h3><p>LSTMs process words one after another.</p><pre>Word1 → Word2 → Word3 → Word4</pre><p>The next word cannot be processed until the previous one is finished.</p><p>This means:</p><ul><li>Slow training</li><li>Poor hardware utilization</li><li>Difficulty scaling to massive datasets</li></ul><h3>2. Long Sentences Were Still Difficult</h3><p>Even with memory cells, understanding relationships across hundreds of words remained challenging.</p><p>Consider:</p><blockquote><em>“The book that I borrowed from the library last month and finally finished yesterday, it was fascinating.”</em></blockquote><p>Connecting “book” and “fascinating” is not trivial when dozens of words separate them.</p><h3>3. Parallelization Was Nearly Impossible</h3><p>Modern GPUs thrive on parallel computations. But RNNs and LSTMs are inherently sequential. This became a major bottleneck as datasets and models grew larger. Researchers needed a completely different approach.</p><h4>The Question That Changed AI</h4><p>Instead of asking:</p><blockquote><em>“How do we improve memory?”</em></blockquote><p>Researchers asked:</p><blockquote><em>“Why process words sequentially at all?”</em></blockquote><p>What if every word could directly look at every other word in the sentence?</p><p>This simple question led to one of the most important breakthroughs in AI — <strong>Transformers</strong></p><h3>Transformer — The Magic of Attention:</h3><p>Suppose we have the sentence:</p><blockquote><em>“Rahul hit the ball because he was confident.”</em></blockquote><p>Who does “he” refer to? Humans immediately know that it refers to “Rahul.” Transformers solve this using a mechanism called <strong>Self-Attention</strong>.</p><p>Every word in the sentence asks:</p><ul><li>Which words are important to me?</li><li>How much attention should I pay to them?</li></ul><p>Instead of passing information step by step, words can directly interact with each other. This completely changes the game.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/761/0*iFwxbhG7a00RT3LL" /></figure><p><strong>Self-Attention Mechanism</strong><br>The self-attention mechanism enables the model to capture dependencies between different positions within a sequence by attending to all positions simultaneously. In the previous blog, we discussed the use of queries and key-value pairs to calculate attention scores. The attention scores determine the importance or relevance of each key-value pair to the given query. Self-attention extends this mechanism to operate within a single sequence without requiring external inputs.</p><h4>Why Attention Was Revolutionary?</h4><p>Consider these two approaches.</p><h4>RNN/LSTM:</h4><pre>Rahul → hit → the → ball → because → he</pre><p>Information must travel through every intermediate word.</p><h4>Transformer:</h4><pre>he ---------------&gt; Rahul</pre><ul><li>A direct connection.</li><li>No long chains.</li><li>No forgotten context.</li><li>No dependency on distance.</li></ul><h4>Parallel Processing: The Superpower</h4><p>Another breakthrough was parallelization. Instead of processing:</p><pre>Word1<br>↓<br>Word2<br>↓<br>Word3</pre><p>Transformers process:</p><pre>Word1<br>Word2<br>Word3<br>Word4</pre><p>All at the same time. This made training:</p><ul><li>Faster</li><li>More efficient</li><li>Scalable to billions of parameters</li></ul><p>Without this capability, modern Large Language Models would simply not exist.</p><h4>From Transformers to Generative AI</h4><p>The Transformer architecture became the foundation for:</p><ul><li>GPT</li><li>ChatGPT</li><li>Gemini</li><li>Claude</li><li>Llama</li></ul><p>These systems can:</p><ul><li>Write articles</li><li>Generate code</li><li>Translate languages</li><li>Answer questions</li><li>Create entirely new content</li></ul><p>All because they learned one powerful idea:</p><blockquote><em>Pay attention to what matters.</em></blockquote><h3>Conclusion:</h3><p>RNNs taught machines to read. LSTMs taught them to remember. But Transformers taught them to <strong>understand relationships and context at scale</strong>.</p><p>The rise of Generative AI isn’t just a story about bigger models or more data. It’s the story of a simple yet profound realization —</p><blockquote><em>Sometimes, intelligence isn’t about remembering everything. It’s about knowing what deserves your attention.</em></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/944/1*C-R7CEgLu1PMsTnhhrFaxQ.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c1a2f9bcd841" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Feature Engineering: The Backbone of an Effective ML Model]]></title>
            <link>https://pawashejagruti.medium.com/feature-engineering-the-backbone-of-an-effective-ml-model-f143aff000f0?source=rss-73530dc3fd0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/f143aff000f0</guid>
            <category><![CDATA[data-preparation]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[feature-engineering]]></category>
            <category><![CDATA[feature-selection]]></category>
            <category><![CDATA[data-preprocessing]]></category>
            <dc:creator><![CDATA[Jagruti Pawashe]]></dc:creator>
            <pubDate>Fri, 17 Apr 2026 07:59:12 GMT</pubDate>
            <atom:updated>2026-04-17T07:59:12.969Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Introduction:</strong></h3><p>In machine learning, it’s easy to believe that better models lead to better results. After all, with advanced techniques like Gradient Boosting and Deep Learning, we have more powerful tools than ever before. But in practice, even the most sophisticated models can fail to deliver if the data they learn from isn’t properly prepared.</p><p><strong>Consider a common scenario:</strong> The model is trained, tuned, and optimized - yet its performance barely improves. The algorithms are sound, the implementation is correct, but something still feels off. More often than not, the issue lies not in the model, but in how the data is represented.</p><p>Raw data rarely comes in a form that machine learning models can fully understand. Dates, categories, and numerical values may hold important patterns, but those patterns are often hidden beneath the surface. Without the right transformation, the model cannot capture them.</p><p>This is where feature engineering becomes critical.</p><p><em>“Feature engineering is the process of transforming raw data into meaningful inputs that better represent the underlying problem. It bridges the gap between data and decision-making, enabling models to uncover patterns, improve accuracy, and generate more reliable predictions.”</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GGS9IcygBYhWOnwWLB46Eg.png" /></figure><h3>Types of Features in Machine Learning:</h3><p>In machine learning, <em>features</em> are the inputs that models use to learn patterns. However, not all features are created equal. Depending on their nature and how they are represented, features can be broadly categorized into different types:</p><h4>1. Numerical features:</h4><p>These are quantitative features represented by numbers.</p><p><strong>Types:</strong></p><ul><li><strong>Discrete</strong>: Countable values (e.g., number of purchases, number of students)</li><li><strong>Continuous</strong>: Measurable values (e.g., salary, temperature)</li></ul><p><strong>Example:</strong></p><ul><li>Age = 25</li><li>Salary = 50,000</li></ul><h4>2. Categorical features:</h4><p>A categorical column can contain only discrete features. For example, a person’s gender is a categorical column because it can take only a few values. Birth month is another example because the values must fall between January and December.</p><p>Categorical variables are further divided into binary and non-binary types. Binary variables have two possible categories, while non-binary features can have multiple categories.</p><p><strong>Types:</strong></p><ul><li><strong>Nominal</strong> (no order): Color (Red, Blue, Green)</li><li><strong>Ordinal</strong> (ordered): Education level (High School &lt; Bachelor &lt; Master)</li></ul><p><strong>Common Techniques:</strong></p><ul><li>One-Hot Encoding</li><li>Label Encoding</li></ul><h4>3. Time-series features:</h4><p>Time-series features represent timely data, such as weekly sales or stock price fluctuations over a year. These help models capture temporal patterns.</p><h4>4. Text Features</h4><p>Text data is unstructured and requires transformation using techniques from Natural Language Processing.</p><p>Examples include product reviews or product description columns in a retail dataset.</p><h3>Feature Engineering Techniques</h3><p>Feature engineering is where raw data is transformed into meaningful signals that machine learning models can learn from effectively.</p><p>Feature engineering offers various powerful techniques for converting raw columns into desirable features.</p><ol><li><strong>Handling missing values</strong><br>Missing values can distort model performance, so handling them properly is crucial. There are two main approaches:</li></ol><ul><li><strong>Imputation:</strong> Imputation is the process of filling missing values using available information. For instance, you can use the mean, mode, and median values to replace missing values.</li><li><strong>Deletion:</strong> This method removes rows with missing values and is most suitable when missing data is less than 10% of the dataset size.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*1rWKZJNjvbEhrXdA.png" /></figure><p><strong>2. Handling Outliers:</strong></p><p>Outliers are data points that deviate significantly from the majority of observations. For instance, in a salary dataset where most values range between 40,000 and 1,00,000, an entry like 4,00,000 or 5,000 would be considered an outlier.</p><p>There are several ways to handle outliers effectively:</p><ul><li><strong>Replacement:</strong> Outliers can be substituted with statistical values such as the minimum or maximum of the dataset to limit their impact.</li><li><strong>Transformations:</strong> Applying transformations like logarithmic or square root scaling can help reduce the influence of extreme values.</li><li><strong>Robust Models:</strong> Certain models, such as Decision Tree, Gradient Boosting, and Ridge Regression, are inherently less sensitive to outliers and can handle them more effectively.</li><li><strong>Removal:</strong> If outliers continue to distort the analysis and other methods are ineffective, removing them from the dataset can be considered as a last resort.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*braUyKkdNLepjHUU" /></figure><h4><strong>3. Encoding Categorical Variables:</strong></h4><p>Machine learning models cannot directly process categorical variables, so they must be converted into numerical representations.</p><ul><li><strong>One-hot encoding:</strong> Each category in a categorical feature is represented as a separate column, with a value of 1 if the category is present in the sample and 0 for all other columns.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*YL2nbEwogUCxUQvz.png" /></figure><h4>4. Feature Scaling:</h4><p>Feature scaling ensures that numerical features lie within a standardized range, preventing some features from dominating the learning process due to their larger values.</p><p>Machine learning models that rely on distance-based calculations (e.g., linear regression, k-nearest neighbors, and neural networks) can be affected when features have vastly different scales.</p><p>For example, consider an employee dataset with the following features:</p><ul><li>Age ranges from 20 to 60</li><li>Income ranges from 30,000 to 150,000</li></ul><p>Since income has much larger values than age, a model might assign more importance to income simply because of its scale, not because it is actually more relevant.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/0*AtHAqulDhuG7XRoG.jpeg" /></figure><h4>5. Creating new features</h4><p>Creating new, meaningful features from existing data provides more logical insights into the model.</p><p>For instance, in a house price prediction dataset, if you have the length and breadth columns separately, you can derive a new feature: area = length * breath, which may directly relate to the target variable, price. Inputting this area feature added to the model simplifies discovering hidden patterns.</p><p><strong>6. Feature selection</strong><br>Feature selection is the process of identifying the most relevant variables in a dataset that contribute to predicting the target variable. The goal is to remove redundant, irrelevant, or noisy features so that the resulting model is more accurate, interpretable, and computationally efficient.</p><p>Choosing the right feature selection method can be the difference between a model that generalizes well and one that fails on unseen data. This article explores three main categories of feature selection methods: Filter, Wrapper, and Embedded methods</p><p><strong>Why Feature Selection Matters?</strong></p><p>Machine learning models are only as good as the data they are fed. Irrelevant features can lead to:</p><ul><li>Longer training times</li><li>Higher risk of overfitting</li><li>Lower interpretability</li><li>Poor generalization to new data</li></ul><p>Feature selection helps address these issues by retaining only the features that are most useful for prediction.</p><h3>1. Filter Methods: Independent Statistical Tests</h3><p><strong>Concept:</strong><br>Filter methods rank features based on their statistical relationship with the target variable, independently of the chosen machine learning model.</p><p><strong>When to Use:</strong><br>When you need a quick, model-agnostic approach to remove obviously irrelevant features before model training.</p><p><strong>Common Techniques:</strong></p><ul><li>Correlation Coefficient — Measures linear relationship between features and the target (Pearson, Spearman).</li><li>Chi-Square Test — Assesses the dependence between categorical features and the target.</li><li>Mutual Information — Measures the amount of information a feature shares with the target.</li><li>Variance Threshold — Removes features with very low variance.</li></ul><p><strong>Advantages:</strong></p><ul><li>Very fast</li><li>Works with any machine learning algorithm</li><li>Good as a preprocessing step</li></ul><p><strong>Limitations:</strong></p><ul><li>Does not consider feature interactions</li><li>May drop features that are useful in combination</li></ul><h3>2. Wrapper Methods: Evaluating Feature Subsets</h3><p><strong>Concept:</strong><br>Wrapper methods evaluate different combinations of features by actually training and testing a model. They treat feature selection as a search problem — finding the subset that yields the best model performance.</p><p><strong>When to Use:</strong><br>When accuracy is more important than computational efficiency, and the dataset is not extremely large.</p><p><strong>Common Techniques:</strong></p><ul><li>Forward Selection — Start with no features, add them one at a time.</li><li>Backward Elimination — Start with all features, remove one at a time.</li><li>Recursive Feature Elimination (RFE) — Recursively remove the least important features.</li></ul><p><strong>Advantages:</strong></p><ul><li>Considers feature interactions</li><li>Often produces highly accurate models</li></ul><p><strong>Limitations:</strong></p><ul><li>Computationally expensive</li><li>Higher risk of overfitting if not cross-validated</li></ul><h3>3. Embedded Methods: Selection During Model Training</h3><p><strong>Concept:</strong><br>Embedded methods perform feature selection as part of the model training process. The algorithm itself determines which features are most important.</p><p><strong>When to Use:</strong><br>When you want to combine model training with feature selection in one step.</p><p><strong>Common Techniques:</strong></p><ul><li>LASSO Regression (L1 Regularization) — Forces some coefficients to zero.</li><li>Decision Trees and Random Forests — Rank features by importance during training.</li><li>Elastic Net — Combines L1 and L2 regularization.</li></ul><p><strong>Advantages:</strong></p><ul><li>More efficient than wrapper methods</li><li>Can capture feature interactions</li><li>Integrates directly with training</li></ul><p><strong>Limitations:</strong></p><ul><li>Model-specific</li><li>Importance scores can be hard to interpret for complex models</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/798/0*w799cmZqxsOo3p5m.png" /></figure><h3>Conclusion</h3><p>Feature engineering is an integral part of building machine learning solutions, allowing you to leverage features most efficiently. The process is carried out by data scientists or ML engineers when dealing with any dataset.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f143aff000f0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Cloud Computing — 101]]></title>
            <link>https://pawashejagruti.medium.com/cloud-computing-101-a1ed0db9a516?source=rss-73530dc3fd0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/a1ed0db9a516</guid>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[cloud-services]]></category>
            <category><![CDATA[google-cloud-platform]]></category>
            <dc:creator><![CDATA[Jagruti Pawashe]]></dc:creator>
            <pubDate>Wed, 07 Jan 2026 07:36:48 GMT</pubDate>
            <atom:updated>2026-01-07T07:36:48.246Z</atom:updated>
            <content:encoded><![CDATA[<h3>Cloud Computing — 101</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/400/1*VpRLrJVL03G79BpiGzClfQ.gif" /></figure><h3>What is Cloud?</h3><p>“The cloud” refers to servers that are accessed over the Internet, and the software and databases that run on those servers. Cloud servers are located in data centers all over the world. By using cloud computing, users and companies do not have to manage physical servers themselves or run software applications on their own machines.</p><p>The cloud enables users to access the same files and applications from almost any device, because the computing and storage take place on servers in a data center, instead of locally on the user&#39;s device. This is why a user can log in to their Instagram account on a new phone after their old phone breaks and still find their old account in place, with all their photos, videos, and conversation history. It works the same way with cloud email providers like Gmail or Microsoft Office 365, and with cloud storage providers like Dropbox or Google Drive.</p><p>For businesses, switching to cloud computing removes some IT costs and overhead: for instance, they no longer need to update and maintain their own servers, as the cloud vendor they are using will do that. This especially makes an impact for small businesses that may not have been able to afford their own internal infrastructure but can outsource their infrastructure needs affordably via the cloud. The cloud can also make it easier for companies to operate internationally, because employees and customers can access the same files and applications from any location.</p><h3>How does Cloud Computing Work?</h3><p>Cloud computing is made possible by a technology called virtualization. Virtualization allows for the creation of a simulated, digital-only “virtual” computer that behaves as if it were a physical computer with its own hardware. The technical term for such a computer is a virtual machine. When properly implemented, virtual machines on the same host are sandboxed from one another, so they do not interact with each other at all. The files and applications from one virtual machine are not visible to the other virtual machines, even though they are on the same physical machine.</p><p>Virtual machines also make more efficient use of the hardware hosting them. By running many virtual machines at once, one server can run many virtual “servers,” and a data center becomes like a whole host of data centers, able to serve many organizations. Thus, cloud providers can offer the use of their servers to far more customers at once than they would be able to otherwise, and they can do so at a low cost.</p><p>Even if individual servers go down, cloud servers in general should always be online and always available. Cloud vendors generally back up their services on multiple machines and across multiple regions.</p><p>Users access cloud services either through a browser or through an app, connecting to the cloud over the Internet — that is, through many interconnected networks — regardless of what device they are using.</p><h3>What are cloud services?</h3><p>The resources available in the cloud are known as “services,” since they are actively managed by a cloud provider. Cloud services include infrastructure, applications, development tools, and data storage, among other products. These services are sorted into several different categories, or <em>service models</em>.</p><h3>What are the main service models of cloud computing?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*E1fQzx3mNh7QPyAz.png" /></figure><p><strong>Software-as-a-Service (SaaS)</strong>: Instead of users installing an application on their device, <a href="https://www.cloudflare.com/learning/cloud/what-is-saas/">SaaS</a> applications are hosted on cloud servers, and users access them over the Internet. SaaS is like renting a house: the landlord maintains the house, but the tenant mostly gets to use it as if they owned it. Examples of SaaS applications include Salesforce, MailChimp, and Slack.</p><p><strong>Platform-as-a-Service (PaaS)</strong>: In this model, companies don’t pay for hosted applications; instead, they pay for the things they need to build their own applications. <a href="https://www.cloudflare.com/learning/serverless/glossary/platform-as-a-service-paas/">PaaS</a> vendors offer everything necessary for building an application, including development tools, infrastructure, and operating systems, over the Internet. PaaS can be compared to renting all the tools and equipment necessary for building a house, instead of renting the house itself. PaaS examples include Heroku and Microsoft Azure.</p><p><strong>Infrastructure-as-a-Service (IaaS)</strong>: In this model, a company rents the servers and storage it needs from a cloud provider. They then use that cloud infrastructure to build their applications. <a href="https://www.cloudflare.com/learning/cloud/what-is-iaas/">IaaS</a> is like a company leasing a plot of land on which they can build whatever they want — but they need to provide their own building equipment and materials. IaaS providers include DigitalOcean, Google Compute Engine, and OpenStack.</p><p>Formerly, SaaS, PaaS, and IaaS were the three main models of cloud computing, and essentially all cloud services fit into one of these categories. However, in recent years, a fourth model has emerged:</p><p><strong>Function-as-a-Service (FaaS)</strong>: <a href="https://www.cloudflare.com/learning/serverless/glossary/function-as-a-service-faas/">FaaS</a>, also known as <a href="https://www.cloudflare.com/learning/serverless/what-is-serverless/">serverless computing</a>, breaks cloud applications down into even smaller components that only run when they are needed. Imagine if it were possible to rent a house one little bit at a time: for instance, the tenant only pays for the dining room at dinner time, the bedroom while they are sleeping, the living room while they are watching TV, and when they are not using those rooms, they don’t have to pay rent on them.</p><p>FaaS or serverless applications still run on servers, as do all these models of cloud computing. But they are called “serverless” because they do not run on dedicated machines, and because the companies building the applications do not have to manage any servers.</p><p>Also, serverless functions scale up, or duplicate, as more people use the application — imagine if the tenant’s dining room could expand on demand when more people come over for dinner!</p><h3>What are the different types of cloud deployments?</h3><p>In contrast to the models discussed above, which define how services are offered via the cloud, these different cloud deployment types have to do with where the cloud servers are and who manages them.</p><p>The most common cloud deployments are:</p><ul><li><strong>Private cloud</strong>: A <a href="https://www.cloudflare.com/learning/cloud/what-is-a-virtual-private-cloud/">private cloud</a> is a server, data center, or distributed network wholly dedicated to one organization.</li><li><strong>Public cloud</strong>: A <a href="https://www.cloudflare.com/learning/cloud/what-is-a-public-cloud/">public cloud</a> is a service run by an external vendor that may include servers in one or multiple data centers. Unlike a private cloud, public clouds are shared by multiple organizations. Using virtual machines, individual servers may be shared by different companies, a situation that is called “<a href="https://www.cloudflare.com/learning/cloud/what-is-multitenancy/">multitenancy</a>” because multiple tenants are renting server space within the same server.</li><li><strong>Hybrid cloud</strong>: <a href="https://www.cloudflare.com/learning/cloud/what-is-hybrid-cloud/">hybrid cloud</a> deployments combine public and private clouds, and may even include on-premises legacy servers. An organization may use its private cloud for some services and its public cloud for others, or it may use the public cloud as backup for its private cloud.</li><li><strong>Multi-cloud</strong>: <a href="https://www.cloudflare.com/learning/cloud/what-is-multicloud/">Multi-cloud</a> is a type of cloud deployment that involves using multiple public clouds. In other words, an organization with a multi-cloud deployment rents virtual servers and services from several external vendors — to continue the analogy used above, this is like leasing several adjacent plots of land from different landlords. Multi-cloud deployments can also be hybrid cloud, and vice versa.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/760/0*bIe8wakUiN-3I9UO.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a1ed0db9a516" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Data Chronicles: Understanding Database Types and Their Roles]]></title>
            <link>https://pawashejagruti.medium.com/data-chronicles-understanding-database-types-and-their-roles-587901576872?source=rss-73530dc3fd0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/587901576872</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[database]]></category>
            <category><![CDATA[sql]]></category>
            <dc:creator><![CDATA[Jagruti Pawashe]]></dc:creator>
            <pubDate>Tue, 25 Nov 2025 12:16:02 GMT</pubDate>
            <atom:updated>2025-11-25T12:16:02.698Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*7ja_vDQ6M6x-80mO.jpg" /></figure><p><strong>What is a Database?</strong></p><p>A database is a structured collection of data that is organized and stored in a way that allows easy access, management, and retrieval of information. It’s essentially a digital repository where data is systematically stored, managed, and updated.</p><p>A database is a collection of data stored in an orderly manner. To run a system efficiently, you’d need an adequate memory of the past and present records that went into and came out of that particular system. The same applies to a business or organization requiring several individuals’ cooperative efforts. For this purpose, large and small companies and organizations like hospitals, schools, and universities utilize a beneficial method of acquiring, assembling, and sharing data in systematic ‘entities’ stored within different databases available. Companies can use different types of databases to meet their unique business needs.</p><p><strong>Why Are Databases Important?</strong></p><p>Databases play a crucial role in modern computing and business for several reasons:</p><blockquote>Data Organization and Management: They provide a structured way to organize and store data. Databases can handle large volumes of information efficiently, making it easier to manage and access data.</blockquote><blockquote>Data Integrity and Security: They offer mechanisms to ensure data integrity, preventing inconsistencies or errors in the stored information. Additionally, databases have built-in security features to protect sensitive data from unauthorized access.</blockquote><blockquote>Data Retrieval and Querying: Databases enable quick and efficient retrieval of specific information using query languages (like SQL). This capability is vital for applications and businesses needing fast access to relevant data.</blockquote><blockquote>Scalability and Performance: Databases can be scaled to accommodate increasing amounts of data without compromising performance. This scalability is crucial as data volumes continue to grow exponentially.</blockquote><blockquote>Support for Multiple Users and Applications: Databases allow multiple users and applications to access and manipulate data concurrently while maintaining data integrity and security.</blockquote><blockquote>Application Development: Databases serve as a backend for numerous applications, from simple web applications to complex enterprise systems, providing a reliable and organized data storage solution.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_Q9V2MzQHSXPixND.png" /></figure><p><strong>Different types of Databases:</strong></p><p>There are several types of databases, each designed to handle specific data storage and retrieval needs.</p><ol><li><strong>Hierarchical Database:-</strong></li></ol><ul><li>In a hierarchical database, data is structured in a tree-like manner.</li></ul><p>Those data are stored in hierarchical (Top-down or Bottom-up) order and</p><p>Also, they are represented by the Parent-child relationship</p><ul><li>In a hierarchical database, parents can have many children, but in the case of children, they have only one parent.</li></ul><p>2. <strong>Network Database:-</strong></p><ul><li>The concept of the network database model allows all the child to have multiple parents and also it helps to solve the more complex models uses many to many relationship</li></ul><p>3. <strong>Object-Oriented Database:-</strong></p><p>In Object Oriented Database data are stored in the form of objects.</p><p>It Maintains a structure which is known as a class and displays the data within that.</p><p>4. <strong>Relational Database:-</strong></p><ul><li>Relational Databases are mostly used in DBMS, because of their flexibility and easy-to-implement nature.</li><li>The model works on Normalizing data in the database</li><li>The relational model stores the data in a fixed structure and manipulates them using SQL</li></ul><blockquote><strong>Note:- </strong>Normalizing data:- Normalizing is the process of organizing the data in order to reduce data redundancy and improve data integrity.</blockquote><p>5. <strong>Non-relational or NoSQL Database:-</strong></p><p>A no SQL database uses a variety of formats, such as documents, graphs, wide columns, etc., which offers excellent flexibility and scalability to a database design.</p><h4>Databases are divided into two major types or categories: Relational or Sequence Databases and Non-relational or Non-sequence databases or No SQL databases. An organization may use them individually or combined, depending on the nature of the data and functionality required.</h4><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=587901576872" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Whisper to the Machine: Mastering Prompt Engineering]]></title>
            <link>https://pawashejagruti.medium.com/whisper-to-the-machine-mastering-prompt-engineering-aa704c60ee4c?source=rss-73530dc3fd0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/aa704c60ee4c</guid>
            <category><![CDATA[generative-ai-tools]]></category>
            <category><![CDATA[prompt-engineering]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Jagruti Pawashe]]></dc:creator>
            <pubDate>Thu, 24 Jul 2025 01:49:16 GMT</pubDate>
            <atom:updated>2025-07-24T03:32:58.361Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/900/1*pzGj5ehwKQfZFtP-SD2H2g.gif" /></figure><h3><strong>Introduction:</strong></h3><p>The way we interact with technology is constantly evolving. One of the most exciting recent advancements is in the realm of artificial intelligence (AI), where machines are trained to think, learn, and even communicate like humans. Among the myriad of developments in areas such as generative AI, there’s a subtle art that’s gaining prominence: prompt engineering.</p><p>Prompt engineering is an artificial intelligence (AI) engineering technique that refines large language models (LLMs) by utilizing specific prompts and recommending targeted outputs. It is also part of the process of refining input to various generative AI (GenAI) services to generate text or images. Prompt engineering enables generative AI tools to create a wide range of content and digital artifacts, including robotic process automation bots, 3D assets, scripts, and robot instructions.</p><p><strong>Why is prompt engineering important?</strong></p><p>Prompt engineering is critical because it directly influences the quality, relevance, and accuracy of generative AI outputs. A well-crafted prompt helps ensure that the AI comprehends the user’s intent and produces meaningful responses, minimizing the need for extensive postprocessing. As gen AI systems become more widely adopted across industries, a prompt engineering guide serves as the key to unlocking their full potential by bridging the gap between raw queries and actionable outputs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/745/1*j8Q9C-2MkXCZMbbSLKvZmw.png" /></figure><h3><strong>Prompt Elements</strong></h3><p>Now we know that prompt is the key to getting the best out of LLMs’ capabilities. To achieve this, the prompt should be clear and precise.</p><p>For example, if we are interested in understanding prompt engineering from an LLM, you might try something like:</p><blockquote>Explain the concept of prompt engineering. Keep the explanation short, only a few sentences, and don’t be too descriptive.</blockquote><p>It’s not clear from the prompt above how many sentences we want or in what style. You might still get some responses, but a better prompt would be the one that is precise. Something like:</p><blockquote>Use 2–3 sentences to explain the concept of prompt engineering to a high school student.</blockquote><p>As we cover more examples of prompt engineering and applications, you will notice that a prompt consists broadly of the following elements:</p><ol><li><strong>Instruction</strong><br>A specific task or instruction you want the LLM to perform. For example, an instruction can be to act like a specific persona — “You are a polite librarian,” or “You are a sarcastic chef.”. A task can be something that you want an LLM to perform — “Summarize the book,” or “Give me a recipe.”</li><li><strong>Context</strong><br>Additional information or background details are provided to guide the LLM in generating more accurate and relevant responses. Including context helps the model better understand the specific circumstances or conditions surrounding a query, enabling it to produce more contextually appropriate answers. This can be particularly useful when dealing with ambiguous or multifaceted questions, as context assists the model in interpreting and responding to the input in a more nuanced and informed manner. The context can be domain-specific information as well.</li><li><strong>Input Data</strong><br>The input or the question that we are interested in is the response. For example, for the “summarize the book” instruction, the input can be “‘The Power of Habit’ by Charles Duhigg.”. To give me a recipe, the input can be “An Italian pizza with pineapple.”</li><li><strong>Output Indicator:</strong></li></ol><p>Format for the output. For Example, you can add “Give bullet points for the recipe” in case of a recipe task.</p><p>If we combine all the above elements for the recipe task, we get the following prompt:</p><blockquote>You are a sarcastic chef. Give me a recipe for an Italian pizza with pineapple. Give bullet points for the recipe.</blockquote><p>The response from LLM is:</p><p>Ah, an Italian pizza with pineapple, you rebel! Here’s a recipe that might raise a few eyebrows among pizza purists:</p><ul><li>Dough Drama: Pretend you’re in Naples while making the pizza dough because nothing says authentic like a bit of theater. Throw the flour around for that extra flair.</li><li>Tomato Tango: Use San Marzano tomatoes because if it’s not from Italy, it’s probably not worth your time. Crush them with your bare hands for that primal connection.</li><li>Cheese Choreography: Mozzarella, of course, but do a little dance while you’re sprinkling it on. Maybe even whisper “cheese amore” for good measure.</li><li>Herb Hilarity: Basil is a must, but don’t just place it gently — give it a dramatic flourish like you’re presenting it on a cooking show. Maybe throw in a few oregano twirls for added mystery.</li><li>Olive Oil Opera: Drizzle some extra virgin olive oil, but do it with the passion of an Italian soap opera. Slow and dramatic, as if you’re confessing your love to the pizza.</li><li>Oven Odyssey: Bake it in a wood-fired oven if you have one because authenticity is key. If not, just imagine the wood-fired aroma while it’s in your regular oven.</li><li>Slice Symphony: Cut it into slices with the precision of a ninja. Each slice should tell a story, preferably one with a twist — like pineapple on an Italian pizza.</li><li>Tasting Theater: As you take a bite, channel your inner food critic. Describe the taste as if you’ve discovered the eighth wonder of the culinary world.</li></ul><p>Remember, with this pizza, you’re not just cooking — you’re performing a culinary masterpiece. Buon appetito, with a side of sarcasm!</p><blockquote>It is not necessary to have all these elements in each prompt. It depends on the task at hand and what elements you want to include in your prompt.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/750/1*hjtG5DhqGzRNAONBUCy7xg.png" /></figure><h3>Prompting Techniques</h3><p>At this stage, it’s clear that making prompts better makes a big difference in getting good results for various tasks. That’s what prompt engineering is all about. While the simple examples were enjoyable, in this part, we dive into fancier prompt engineering tricks that help us do cooler and more interesting things. These are called prompting techniques. We will cover a few of the most used prompting techniques here.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/744/1*G2kAPJSiE8X18RlOM5GjlQ.png" /></figure><ol><li><strong>Zero-Shot Prompting:</strong></li></ol><p>No examples given — rely on model’s general knowledge</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*85zgZOpDtEVD0IXtTVHmHw.png" /></figure><p><strong>2. Few-Shot Prompting:</strong></p><p>Provide 1–3 examples to guide the model’s behavior</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/803/1*MAU3v6Z2B_rzZ49TUDCbGw.png" /></figure><p><strong>3. Chain-of-Thought (CoT) Prompting:</strong></p><p>Ask the model to “think step-by-step” before answering</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/665/1*d3SfsRGqshJtNPMdTag0kA.png" /></figure><p><strong>4. Instruction Prompting:</strong></p><p>Directly tell the model what to do in plain language.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/862/1*iuZ_e1hjKLqag-_mwKzupA.png" /></figure><p><strong>5. Role Prompting (Persona Prompt):</strong></p><p>Assign the model a specific role.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*n8XRk5bMPoMMs8KtFkcJfg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/741/1*L65TeR599A59ziqY_41ZyQ.png" /></figure><h3>Conclusion:</h3><p>The realm of artificial intelligence is vast, intricate, and ever-evolving. As we’ve journeyed through the intricacies of prompt engineering, it’s evident that this field is more than just a technical endeavor — it’s a bridge between human intent and machine understanding. It’s the subtle art of asking the right questions to elicit the desired answers.</p><p>Prompt engineering, though a relatively new discipline, holds the key to unlocking the full potential of AI models, especially Large Language Models. As these models become increasingly integrated into our daily lives, the importance of effective communication cannot be overstated. Whether it’s a voice assistant helping with daily tasks, a chatbot providing customer support, or an AI tool aiding researchers, the quality of interaction hinges on the prompts that guide them.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=aa704c60ee4c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Behind the Data Curtain: Unveiling Key Roles in Data Science]]></title>
            <link>https://pawashejagruti.medium.com/behind-the-data-curtain-unveiling-key-roles-in-data-science-02e7a8371c9e?source=rss-73530dc3fd0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/02e7a8371c9e</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[analyst]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[roles-and-responsibility]]></category>
            <dc:creator><![CDATA[Jagruti Pawashe]]></dc:creator>
            <pubDate>Fri, 04 Oct 2024 05:10:32 GMT</pubDate>
            <atom:updated>2024-10-04T09:19:41.816Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*vwp8g1WtaY13ku33.gif" /></figure><p>In today’s data-driven world, “data science” often conjures images of complex algorithms, intricate models, and vast oceans of data. But who are the masterminds behind these cutting-edge technologies? What roles do they play in transforming raw data into actionable insights that drive businesses, innovate solutions, and shape our future?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/432/0*8mM-iKDAJVoYURXr.png" /></figure><h3>Data Scientist — The Magician of Metrics</h3><p>Data scientists use data to create insights and drive business decisions. They use various tools and techniques to collect, process, and analyze data, then present their findings to help organizations make strategic decisions.</p><p>Data scientists determine the questions their team should ask and figure out how to answer them using data. They often develop predictive models for theorizing and forecasting.</p><p>A data scientist might do the following tasks on a day-to-day basis:</p><ul><li>Find patterns and trends in datasets to uncover insights.</li><li>Create algorithms and data models to forecast outcomes.</li><li>Use machine learning techniques to improve the quality of data or product offerings.</li><li>Communicate recommendations.</li><li>Stay on top of innovations in the data science field.</li></ul><blockquote>Skills Required — Python/R, SQL, Machine Learning, Deep Learning, Statistics, Cloud Computing, Visualization tools, Big Data, Github, Model Deployment, etc.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*nu3k4bbOPGULbcYi.png" /></figure><h3><strong>Data Analyst — The Detective of Data</strong></h3><p>A data analyst is a professional who collects, analyzes, and interprets data to help organizations make decisions. They work with data from various sources, including company databases and external sources, to identify insights that can help businesses maximize the value of their data.</p><p>Generally speaking, though, the tasks data analysts must perform daily include the following:</p><ul><li>Designing and maintaining data systems and databases; this includes fixing coding errors and other data-related problems.</li><li>Mining data from primary and secondary sources, then reorganizing the data in a format that can be easily read by either humans or machines.</li><li>Using statistical tools to interpret datasets, paying particular attention to trends and patterns that could be valuable for diagnostic and predictive analytics efforts.</li><li>Demonstrating the significance of their work in the context of local, national, and global trends that impact both their organization and industry.</li><li>Preparing reports for executive leadership that effectively communicate trends, patterns, and predictions using relevant data.</li><li>Collaborating with programmers, engineers, and organizational leaders to identify opportunities for process improvements, recommend system modifications, and develop policies for data governance.</li><li>Creating appropriate documentation that allows stakeholders to understand the steps of the data analysis process and duplicate or replicate the analysis if necessary.</li></ul><blockquote>Skills Required — Data Visualization tools, Excel, SQL, Python/R, SAS, Statistics, etc.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*1TM7eYypMrZ7nq1A.jpg" /></figure><h3>Data Engineer — The Builder of Data Highways</h3><p>A data engineer integrates, transforms, and consolidates data from various structured and unstructured data systems into structures that are suitable for building analytics solutions. The data engineer also helps design and support data pipelines and data stores that are high-performing, efficient, organized, and reliable, given a specific set of business requirements and constraints.</p><p>These are some common tasks you might perform when working with data:</p><ul><li>Acquire datasets that align with business needs</li><li>Support the development of data streaming systems</li><li>Implement new systems for data analytics and business intelligence operations</li><li>Develop business intelligence reports for company advisors</li><li>Develop algorithms to transform data into useful, actionable information</li><li>Build, test, and maintain database pipeline architectures</li><li>Collaborate with management to understand company objectives</li><li>Create new data validation methods and data analysis tools</li><li>Ensure compliance with data governance and security policies</li></ul><blockquote>Skills Required — Cloud Services, SQL/ No SQL, Big Data, etc</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*WfCjtUMo0oKU2JHa.png" /></figure><h3>Business Analyst — The Interpreter of Business Needs</h3><p>Business analysts use data to form insights and recommend business and other organizational changes. Business analysts identify areas that can be improved to increase efficiency and strengthen business processes. They often work closely with others throughout the organization’s hierarchy to communicate their findings and help implement changes.</p><p>Tasks and duties can include:</p><ul><li>Identifying and prioritizing the organization’s functional and technical needs and requirements.</li><li>Using SQL and Excel to analyze large data sets.</li><li>Compiling charts, tables, and other elements of data visualization.</li><li>Creating financial models to support business decisions.</li><li>Understanding business strategies, goals, and requirements.</li><li>Planning enterprise architecture (the structure of a business).</li><li>Forecasting, budgeting, and performing variance and financial analysis.</li></ul><blockquote>Skills Required — SQL, Python, Visualization tools, Excel, etc.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/602/0*sHgi86974rZKFmlS.png" /></figure><h3>ML Engineer — The Engineer of Learning</h3><p>A machine learning engineer (ML engineer) is a programmer who designs and builds software that can automate artificial intelligence and machine learning (AI/ML) models.</p><p>ML engineers build large-scale systems that take in massive data sets and use them to train algorithms that can learn cognitive tasks and generate useful insights and predictions.</p><p>Machine learning engineers manage the entire data science pipeline, including sourcing and preparing data, building and training models, and deploying models to production.</p><p>To achieve these and related tasks, machine learning engineers perform the following activities in an organization:</p><ul><li>Analyze big datasets and then determine the best method to prepare the data for analysis.</li><li>Ingest source data into machine learning systems to enable machine learning training.</li><li>Collaborate with other data scientists and build effective data pipelines.</li><li>Build the infrastructure required to deploy a machine-learning model in production.</li><li>Manage, maintain, scale, and improve machine learning models already running in production environments.</li><li>Work with common ML algorithms and relevant software libraries.</li><li>Optimize and tweak ML models according to how they behave in production.</li><li>Communicate with relevant stakeholders and key users to understand business requirements, and also clearly explain the capabilities of the ML model.</li><li>Deploy models to production, initially as a prototype, and then as an API that can serve predictions for end users.</li></ul><blockquote>Skills Required — Machine Learning, Deep Learning, Model Deployment, Python/R, Cloud Services, etc.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/0*fH29tovwXlUk7cZA.jpg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=02e7a8371c9e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Dynamic duo of Python & SQL]]></title>
            <link>https://pawashejagruti.medium.com/the-dynamic-duo-of-python-sql-1dd353d19a4c?source=rss-73530dc3fd0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/1dd353d19a4c</guid>
            <category><![CDATA[sql]]></category>
            <category><![CDATA[sqlite]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[pandas]]></category>
            <dc:creator><![CDATA[Jagruti Pawashe]]></dc:creator>
            <pubDate>Sat, 16 Sep 2023 00:42:01 GMT</pubDate>
            <atom:updated>2023-09-17T04:39:22.119Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*VdQlBAFEuCHk9XWS.png" /></figure><p><strong>Why a need to connect SQL &amp; Python?</strong></p><p>In today’s data-driven world, information is power. In the ever-expanding universe of data, the ability to extract meaningful insights has become the North Star guiding businesses, organizations, and individuals toward informed decision-making.</p><p>Python is a general-purpose programming language that is known for its simplicity and readability. Python is also very versatile, and it can be used for a wide variety of tasks, including data analysis, machine learning, and web development.</p><p>SQL stands for Structured Query Language. It is a declarative language, which means that you tell the database what you want to do, but not how to do it. SQL is a very powerful language, and it can be used to perform complex queries on large datasets.</p><p>Python, celebrated for its versatility and simplicity, has firmly established itself as a beloved tool in the arsenal of data analysts and scientists. But what makes Python truly shine in the world of data is its symbiotic relationship with Structured Query Language (SQL), the key that unlocks the doors to vast data repositories.</p><p>Connecting Python to SQL databases allows you to leverage the power of Python for data analysis, manipulation, and visualization while using SQL to interact with and manage your data in a database.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*DsIpnvUFCtKFEXCWLx3g5Q.gif" /></figure><p>There are various ways to establish connectivity between Python and SQL databases, with the most common methods being:</p><p><strong>Using Database Libraries:</strong></p><ul><li><strong>SQLite3 (built-in):</strong> Python comes with a built-in SQLite3 library, which allows you to interact with SQLite databases without any additional installations.</li><li><strong>Third-party Libraries:</strong> For other relational databases like MySQL, PostgreSQL, Oracle, or SQL Server, you can use third-party libraries like:</li><li>mysql-connector-python or pymysql for MySQL.</li><li>psycopg2 for PostgreSQL.</li><li>cx_Oracle for Oracle.</li><li>pyodbc for SQL Server.</li></ul><p>These libraries provide Python functions and classes to establish connections, execute SQL queries, and fetch results from databases.</p><h3>Connecting to the Database</h3><p>Connecting to the SQLite Database can be established using the <strong>connect()</strong> method, passing the name of the database to be accessed as a parameter. If that database does not exist, then it’ll be created.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*YI_22yJ8Qi2XarFC.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*hVPvlUS9iyWUk7nI.gif" /></figure><pre>sqliteConnection = sqlite3.connect(&#39;sql.db&#39;)</pre><p>But what if you want to execute some queries after the connection is made? For that, a cursor has to be created using the<strong> cursor()</strong> method on the connection instance, which will execute our SQL queries.</p><pre>cursor = sqliteConnection.cursor()</pre><p>Creating an SQLite engine:</p><pre>engine = create_engine(&quot;sqlite:///sql.db&quot;)</pre><h4>PostgreSQL</h4><p>The PostgreSQL dialect uses psycopg2 as the default DBAPI. Other PostgreSQL DBAPIs include pg8000:</p><pre># default<br>engine = create_engine(&quot;postgresql://scott:tiger@localhost/mydatabase&quot;)</pre><pre># psycopg2<br>engine = create_engine(&quot;postgresql+psycopg2://scott:tiger@localhost/mydatabase&quot;)</pre><pre># pg8000<br>engine = create_engine(&quot;postgresql+pg8000://scott:tiger@localhost/mydatabase&quot;)</pre><h4>MySQL</h4><p>The MySQL dialect uses MySQL client as the default DBAPI. There are other MySQL DBAPIs available, including PyMySQL:</p><pre># default<br>engine = create_engine(&quot;mysql://root:password@localhost/database_name&quot;)</pre><pre># mysqlclient (a maintained fork of MySQL-Python)<br>engine = create_engine(&quot;mysql+mysqldb://root:password@localhost/db_name&quot;)</pre><pre># PyMySQL<br>engine = create_engine(&quot;mysql+pymysql://root:password@localhost/db_name&quot;)</pre><h4>Oracle</h4><p>The Oracle dialect uses cx_oracle as the default DBAPI:</p><pre>engine = create_engine(&quot;oracle://root:password@127.0.0.1:1521/db_name&quot;)</pre><pre>engine = create_engine(&quot;oracle+cx_oracle://root:password@tnsname&quot;)</pre><h4>Microsoft SQL Server</h4><p>The SQL Server dialect uses pyodbc as the default DBAPI. pymssql is also available:</p><pre># pyodbc<br>engine = create_engine(&quot;mssql+pyodbc://root:password@mydsn&quot;)</pre><pre># pymssql<br>engine = create_engine(&quot;mssql+pymssql://root:password@hostname:port/dbname&quot;)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1_HPpDkN2ujjmkUqGu4egw.gif" /><figcaption>SQL + PYTHON Integration</figcaption></figure><p><strong>Pandas function for effective &amp; easy working of SQL queries:</strong></p><p><strong>SQL and pandas both have a place in a functional data analysis tech stack</strong>, and today we’re going to look at how to use them both together most effectively. We suggested doing the really heavy lifting directly in the database instance via SQL, then doing the finer-grained data analysis on your local machine using pandas — but we didn’t actually go into how you could do that.</p><p><strong>pd.read_sql:</strong> To read SQL table into a DataFrame using only the table name, without executing any query we use the <strong>read_sql_table()</strong> method in Pandas. This function does not support DBAPI connections.</p><blockquote><em>pandas.read_sql_table(table_name, con, schema=None, index_col=None, coerce_float=True, parse_dates=None, columns=None, chunksize=None)</em></blockquote><p><strong>pd.read_sql:</strong> Returns a DataFrame corresponding to the result set of the query string.</p><blockquote>pandas.read_sql_query(<em>sql_query</em>, <em>con</em>, <em>index_col=None</em>, <em>coerce_float=True</em>, <em>params=None</em>, <em>parse_dates=None</em>, <em>chunksize=None</em>, <em>dtype=None</em>, <em>dtype_backend=_NoDefault.no_default</em>)</blockquote><p><strong>data.to_sql:</strong> The to_sql() function is used to write records stored in a DataFrame to a SQL database.</p><blockquote>DataFrame.to_sql(<em>name</em>, <em>con</em>, <em>*</em>, <em>schema=None</em>, <em>if_exists=’fail’</em>, <em>index=True</em>, <em>index_label=None</em>, <em>chunksize=None</em>, <em>dtype=None</em>, <em>method=None</em>)</blockquote><p><strong>pd.read_sql: </strong>pandas read_sql() the function is used to read SQL queries or database tables into DataFrame. This is a wrapper read_sql_query() and read_sql_table() functions, based on the input it calls these functions internally and returns the SQL table as a two-dimensional data structure with labeled axes.</p><blockquote>pandas.<strong>read_sql</strong>(<em>sql_query_or_table</em>, <em>con</em>, <em>index_col=None</em>, <em>coerce_float=True</em>, <em>params=None</em>, <em>parse_dates=None</em>, <em>columns=None</em>, <em>chunksize=None</em>, <em>dtype_backend=_NoDefault.no_default</em>, <em>dtype=None</em>)</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1dd353d19a4c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Joins]]></title>
            <link>https://pawashejagruti.medium.com/joins-b667353f0f2c?source=rss-73530dc3fd0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/b667353f0f2c</guid>
            <category><![CDATA[data]]></category>
            <category><![CDATA[dbms]]></category>
            <category><![CDATA[sql]]></category>
            <category><![CDATA[joins-in-sql]]></category>
            <dc:creator><![CDATA[Jagruti Pawashe]]></dc:creator>
            <pubDate>Fri, 01 Sep 2023 04:22:35 GMT</pubDate>
            <atom:updated>2026-09-12T06:09:59.382Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*0Xr0EEjdEb5jprT2.jpeg" /></figure><p><strong>SQL Join</strong> statement is used to combine data or rows from two or more tables based on a common field between them. Different types of Joins are as follows:</p><ul><li>INNER JOIN</li><li>LEFT JOIN</li><li>RIGHT JOIN</li><li>FULL JOIN</li><li>NATURAL JOIN</li><li>CROSS JOIN</li><li>SELF JOIN</li></ul><p>1. INNER JOIN :</p><p>The INNER JOIN keyword selects all rows from both tables as long as the condition is satisfied. This keyword will create the result-set by combining all rows from both the tables where the condition satisfies i.e. the value of the common field will be the same.</p><p>Consider the two tables below as follows:</p><p><strong>Student</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/0*av6Cn_Pfm99nh7QU.png" /><figcaption><strong>Student’s Table</strong></figcaption></figure><p><strong>Course</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/361/0*X5-MMjMoreY7Kb8Z.png" /><figcaption><strong>Course Table</strong></figcaption></figure><pre>SELECT table1.column1,table1.column2,table2.column1,....<br>FROM table1 <br>INNER JOIN table2<br>ON table1.matching_column = table2.matching_column;</pre><pre>SELECT StudentCourse.COURSE_ID, Student.NAME, Student.AGE FROM Student<br>INNER JOIN StudentCourse<br>ON Student.ROLL_NO = StudentCourse.ROLL_NO;</pre><p>OUTPUT:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/546/0*kLJPXKYxuEQeH3m9.png" /></figure><p>2. LEFT JOIN:</p><p>This join returns all the rows of the table on the left side of the join and matches rows for the table on the right side of the join. For the rows for which there is no matching row on the right side, the result set will contain <em>null</em>. LEFT JOIN is also known as LEFT OUTER JOIN.</p><pre>SELECT table1.column1,table1.column2,table2.column1,....<br>FROM table1 <br>LEFT JOIN table2<br>ON table1.matching_column = table2.matching_column;</pre><pre>SELECT Student.NAME,StudentCourse.COURSE_ID <br>FROM Student<br>LEFT JOIN StudentCourse <br>ON StudentCourse.ROLL_NO = Student.ROLL_NO;</pre><p>OUTPUT:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/0*k1UNGmjfjuZmwfyu.png" /></figure><p>3. RIGHT JOIN:</p><p>The <strong>Right Join</strong> or <strong>Right Outer Join</strong> query in SQL returns all rows from the right table, even if there are no matches in the left table. This means that if the ON clause matches 0 (zero) records in the left table; the join will still return a row in the result, but with a NULL value in each column of the left table.</p><pre>SELECT table1.column1,table1.column2,table2.column1,...<br>FROM table1 <br>RIGHT JOIN table2<br>ON table1.matching_column = table2.matching_column;</pre><pre>SELECT Student.NAME,StudentCourse.COURSE_ID <br>FROM Student<br>RIGHT JOIN StudentCourse <br>ON StudentCourse.ROLL_NO = Student.ROLL_NO;</pre><p>OUTPUT:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/361/0*SzqH6CtGJOqpHzwY.png" /></figure><p>4. FULL JOIN:</p><p><strong>SQL Full Join</strong> creates a new table by joining two tables as a whole. The joined table contains all records from both tables and fills in NULLs for missing matches on either side. In short, a full join is a type of outer join that combines the results of both left and right joins.</p><pre>SELECT table1.column1,table1.column2,table2.column1,....<br>FROM table1 <br>FULL JOIN table2<br>ON table1.matching_column = table2.matching_column;</pre><pre>SELECT Student.NAME,StudentCourse.COURSE_ID <br>FROM Student<br>FULL JOIN StudentCourse <br>ON StudentCourse.ROLL_NO = Student.ROLL_NO;</pre><p>5. NATURAL JOIN:</p><p>Natural join can join tables based on the common columns in the tables being joined. A natural join returns all rows by matching values in common columns having the same name and data type of columns and that column should be present in both tables.</p><p>Both tables must have at least one common column with the same column name and the same data type.</p><p>The two tables are joined using cross-join.</p><p>DBMS will look for a common column with the same name and data type Tuples having exactly the same values in common columns are kept in the result.</p><p>6. CROSS JOINS:</p><p>The CROSS JOIN is used to generate a paired combination of each row of the first table with each row of the second table. This join type is also known as cartesian join.</p><p>Suppose that we are sitting in a coffee shop and we decide to order breakfast. Shortly, we will look at the menu and we will start thinking of which meal and drink combination could be tastier. Our brain will receive this signal and begin to generate all meal and drink combinations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/585/1*QM5MKfl39Bfhd8bDJA-ntg.jpeg" /></figure><p>7. SELF JOIN:</p><p>The SELF JOIN in SQL, as its name implies, is used to join a table to itself. This means that each row in a table is joined to itself and every other row in that table. However, referencing the same table more than once within a single query will result in an error. To avoid this, SQL SELF JOIN aliases are used.</p><pre>SELECT<br>    column_names<br>FROM<br>    Table1 t1<br>[INNER | LEFT]  JOIN Table1 t2 ON<br>    join_predicate;</pre><p>You can also create the SELF JOIN with the help of the WHERE clause.</p><pre>SELECT <br>    column_names<br>FROM <br>    Table1 t1, Table1 t2<br>WHERE <br>    condition;</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/963/0*jGqLiTQg8I-bRiBE.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b667353f0f2c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Let’s Understand the Data…]]></title>
            <link>https://pawashejagruti.medium.com/lets-understand-the-data-c76aa8cf3905?source=rss-73530dc3fd0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/c76aa8cf3905</guid>
            <category><![CDATA[statistics]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[data-visualization]]></category>
            <category><![CDATA[data-storytelling]]></category>
            <dc:creator><![CDATA[Jagruti Pawashe]]></dc:creator>
            <pubDate>Mon, 14 Aug 2023 04:21:10 GMT</pubDate>
            <atom:updated>2023-08-27T10:59:48.095Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/647/0*xyNG3AciaJkVTedG.jpg" /></figure><p><strong>What is Data?</strong></p><p>Data is raw pieces of facts/figures that need to be processed further to gain some valuable insights from it. As it is observed nowadays, “Data has become the fuel of the future”.</p><p>Data is the raw material of the 21st century. It can be used to power everything from businesses to governments to social movements. But data is not created equal. The data that is valuable for one domain or problem may be useless for another. This is why it is important to understand the different types of data and how they can be used.</p><p><strong>Types of Data-</strong></p><p>Data can be classified into different types based on various characteristics &amp; properties. These classifications help us to understand the nature of the data and determine valuable analysis and interpretation.</p><ol><li><strong>Numerical Data:</strong></li></ol><p>Numerical data, also known as quantitative data, is data that you typically present in number form, and it doesn’t include any language or descriptive form. It’s always measurable, and you can add it together. You can perform mathematical and arithmetical operations on it, and you can express it in decimal or fraction form.</p><p>a. Discrete Data: Discrete data represents whole numbers or counts that have a finite or countable number of values.For e.g. Number of Bank accounts, Number of cars, etc.</p><p>b. Continuous Data: Continuous data represents measurements or values that can take any value within a certain range.For e.g. Height, weight, distance, etc.</p><p><strong>2. Categorical Data:</strong></p><p>Categorical data represents qualitative variables that can be grouped into specific categories or classes.</p><p>a. Nominal Data: Nominal data consists of categories that do not have any specific order or ranking. These categories can simply be labels or names. E.g. Car Brands(Toyota, BMW, Audi, etc.), Hair color(Brown, Black, Blonde, etc.), Marital status ( Single, Married, Divorced), etc</p><p>b. Ordinal Data: Ordinal data represents categories with a natural order or ranking. The categories have a relative position or value. E.g. Educational level(Primary, Secondary, College) or food satisfaction ratings ( poor, average, good, excellent)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Cc5a3FV-mDU4-Lamy0ucRA.png" /></figure><p><strong>3. Dichotomous Data:</strong></p><p>Dichotomous or Binary data occur when you can place an observation into only two categories. It tells you that an event occurred or that an item has a particular characteristic. For instance, an inspection process produces binary pass/fail results. Or, when a customer enters a store, there are two possible outcomes — sale or no sale. Binary data is commonly used in fields such as computer science and statistics to represent logical or boolean variables.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/576/0*5kX_ivuSIBMl-gvm" /></figure><p><strong>4. Textual Data:</strong></p><p>Textual data refers to unstructured data in the form of textual information such as sentences, paragraphs, or Word documents. Textual data often requires Natural Language Processing techniques to extract meaningful insights or perform analysis</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/837/0*NRo9-N3q440LxAXu.png" /></figure><p><strong>5. Temporal Data:</strong></p><p>Temporal data refers to information that changes over time. It includes data points associated with specific timestamps or time intervals. Temporal data is most commonly used in finance, weather forecasting as well and stock market analysis.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/698/1*THd-85zv9bopGsb4ek7aYA.png" /></figure><p><strong>6. Spatial data:</strong></p><p>Spatial data is any type of data that directly or indirectly references a specific geographical area or location. Sometimes called geospatial data or geographic information, spatial data can also numerically represent a physical object in a geographic coordinate system. However, spatial data is much more than a spatial component of a map. Analyzing this data provides a better understanding of how each variable impacts individuals, communities, populations, etc.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/448/0*_Rqu-N65AbEvfpyU.png" /></figure><p><strong>7. Image &amp; Video Data:</strong></p><p>Image and video data are visual information captured in the form of images or sequences of images (called frames) captured and eventually displayed at a given frequency. By stopping at a specific frame of the sequence, a single video frame, i.e. an image, is obtained. Image and video analytics applications utilizing deep learning algorithm deployment have long surpassed human abilities to recognize data patterns in images and videos.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/884/0*5oJAjnq1gwSC8_It.png" /></figure><p><strong>8. Speech Data:</strong></p><p>Any digital information with speech or music stored on and played through a system is known as an audio file or sound file. Gathering and recording sounds, either for human enjoyment or for use in automated learning systems, is known as “audio data collection”. For example, voice recognition systems rely heavily on audio data, which has risen in prominence in recent years for usage in AI and ML systems.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*D-FN5OdXQEJ6WTCt.jpg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c76aa8cf3905" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Quantum- A New World for Computing…]]></title>
            <link>https://pawashejagruti.medium.com/quantum-a-new-world-for-computing-7015486bd884?source=rss-73530dc3fd0f------2</link>
            <guid isPermaLink="false">https://medium.com/p/7015486bd884</guid>
            <category><![CDATA[quantum]]></category>
            <category><![CDATA[quantum-computing]]></category>
            <category><![CDATA[computing]]></category>
            <category><![CDATA[quantum-physics]]></category>
            <category><![CDATA[science]]></category>
            <dc:creator><![CDATA[Jagruti Pawashe]]></dc:creator>
            <pubDate>Wed, 22 Mar 2023 14:42:17 GMT</pubDate>
            <atom:updated>2023-03-22T14:42:17.051Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*V2GoRP5Pz7BOrfGZ.jpg" /></figure><p><strong>Introduction:</strong></p><p>Quantum computing is a new way of processing information that relies on the principles of quantum physics. In traditional computing, data is represented using bits that can be either a 0 or a 1. However, in quantum computing, information is represented using quantum bits or qubits, which can be in multiple states simultaneously.</p><p>This means that quantum computers can process a vast amount of information in parallel, making them much faster than classical computers for specific tasks. For example, quantum computers are particularly well-suited for solving complex mathematical problems and for complex simulating systems.</p><p><strong>What is Quantum in Quantum Computing?</strong></p><p>In quantum computing, the term “quantum” refers to the fundamental unit of quantum information, called a qubit. A qubit is a quantum system that can be in multiple states at the same time, as opposed to classical bits, which can only be in one of two states (0 or 1) at any given time.</p><p>The term “quantum” comes from the fact that qubits are subject to the laws of quantum mechanics, which describe the behavior of matter and energy at the smallest scales. In classical physics, particles like electrons and photons are treated as individual objects with specific properties, such as position and momentum, that can be precisely measured. In quantum mechanics, however, particles are described by wave functions that give the probability of finding the particle in a particular state.</p><p>The behavior of qubits is based on two fundamental principles of quantum mechanics: superposition and entanglement. Superposition allows qubits to be in multiple states at the same time, which is essential for the parallel processing capabilities of quantum computers. Entanglement allows qubits to become correlated in such a way that the state of one qubit is dependent on the state of the other, which is important for performing certain quantum algorithms.</p><p>The study of quantum mechanics and the development of quantum computing is a rapidly evolving field with many challenges still to overcome. However, the potential benefits of quantum computing are significant, with applications in areas such as cryptography, optimization, and simulation.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Rqa12XqUrBkM2A5q.png" /></figure><p><strong>How does this Quantum Computing Work?</strong></p><p>Quantum computing is based on the principles of quantum mechanics, which is a branch of physics that describes the behavior of matter and energy at the smallest scales. In classical computing, information is represented using bits that can be either a 0 or a 1. However, in quantum computing, information is represented using qubits, which can be in multiple states simultaneously.</p><p>This is possible because of a phenomenon called superposition, where a qubit can exist in multiple states at the same time. For example, a qubit can be in a state where it is both a 0 and a 1 at the same time. This allows quantum computers to perform many calculations simultaneously, making them much faster than classical computers for certain tasks.</p><p>In addition to superposition, another important principle in quantum computing is entanglement. Entanglement is a phenomenon where two qubits become correlated in such a way that the state of one qubit is dependent on the state of the other. This can be used to perform computations that are not possible with classical computers.</p><p>To perform a computation on a quantum computer, the qubits are manipulated using quantum gates. These gates are similar to the logic gates used in classical computing but are designed to work with qubits. By applying a series of gates to the qubits, quantum algorithms can be executed, solving problems that are difficult or impossible to solve with classical algorithms.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*go4sH0KwvlJ4r_Fi.jpg" /></figure><p><strong>How is Quantum Computing different from Classical Computers?</strong></p><p>Quantum computing is fundamentally different from classical computing in several ways.</p><ol><li>Information representation: In classical computing, information is represented using bits that can be either a 0 or a 1. In quantum computing, information is represented using qubits, which can be in multiple states simultaneously. This allows quantum computers to perform many calculations simultaneously, making them much faster than classical computers for certain tasks.</li><li>Parallelism: Quantum computers can perform many calculations at the same time, thanks to superposition, which allows a qubit to exist in multiple states simultaneously. This enables quantum computers to solve certain problems much faster than classical computers, which can only perform one calculation at a time.</li><li>Entanglement: In classical computing, bits are independent of each other, and the state of one bit does not affect the state of another bit. In quantum computing, however, qubits can become entangled, which means that the state of one qubit is dependent on the state of another qubit. This allows quantum computers to perform computations that are not possible with classical computers.</li><li>Algorithms: Quantum computing algorithms are fundamentally different from classical computing algorithms. Quantum algorithms take advantage of the unique properties of qubits, such as superposition and entanglement, to perform computations that are not possible with classical algorithms.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*FmIV9YZknb5_9f8L.png" /></figure><p>Despite these advantages, quantum computing is still in its early stages, and there are many challenges that need to be overcome before it becomes a practical technology. However, many researchers are working to develop quantum computers that could revolutionize computing and enable new discoveries in fields like chemistry, cryptography, and artificial intelligence.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7015486bd884" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>