<?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 Suchendra Charith Gondi on Medium]]></title>
        <description><![CDATA[Stories by Suchendra Charith Gondi on Medium]]></description>
        <link>https://medium.com/@sc.gondi2005?source=rss-485c1dfb243a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*UOCn82Xr2CwehNS-</url>
            <title>Stories by Suchendra Charith Gondi on Medium</title>
            <link>https://medium.com/@sc.gondi2005?source=rss-485c1dfb243a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 05:06:39 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sc.gondi2005/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[Linear Regression Explained: A Simple Introduction for Beginners]]></title>
            <link>https://medium.com/@sc.gondi2005/linear-regression-explained-a-simple-introduction-for-beginners-aacf994c4d88?source=rss-485c1dfb243a------2</link>
            <guid isPermaLink="false">https://medium.com/p/aacf994c4d88</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[linear-regression]]></category>
            <category><![CDATA[classification]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Suchendra Charith Gondi]]></dc:creator>
            <pubDate>Tue, 22 Jul 2025 14:57:33 GMT</pubDate>
            <atom:updated>2025-07-22T14:57:33.434Z</atom:updated>
            <content:encoded><![CDATA[<p>Today let us learn the about <strong>Linear Regression</strong> which comes under <strong>Supervised Learning</strong> which is one of the part of <strong>Machine Learning</strong>.</p><pre>Supervised learning is a type of machine learning where an algorithm <br>is trained on a dataset that has been labeled by humans. In this process, <br>the algorithm learns from the labeled input and output data to make <br>predictions or decisions on new, unlabeled data.</pre><p>The topics which we are gonna cover today is:</p><pre>1. What is Linear Regression?<br>2. What is training and testing data? How to divide them?<br>3. What are input features and target variable?<br>4. What are parameters?<br>5. What is loss function and cost function?<br>6. What is gradient descent? How to apply it?<br>7. How to apply linear regression with multiple features?<br>8. What is vectarization?</pre><p>Let us know start with the basic definition,</p><h3><strong>1. Linear Regression:</strong></h3><p>Linear regression is a fundamental statistical method used to model the relationship between a dependent variable and one or more independent variables. It aims to find the best-fitting straight line through the data points, which can then be used to make predictions.</p><p>The core idea of linear regression is to represent the relationship between variables with a linear equation. This equation can be simple, involving one independent and one dependent variable, or multiple, with one dependent variable and two or more independent variables.</p><p>Let us know take an example of house sizes and compare it with the price of house.</p><figure><img alt="A graph of house size vs price of the size, the red dots are the data points which are taken from the official data." src="https://cdn-images-1.medium.com/max/272/1*ls17Vx2DIw233krrqcfJww.png" /><figcaption>This image was taken from Google.</figcaption></figure><h3><strong>2. What is training and testing data? How to divide them?</strong></h3><p>In machine learning, <strong>training data</strong> is the portion of a dataset used to “teach” or fit a model. The model learns patterns, relationships, and features from this data. <strong>Testing data</strong> is a separate, unseen portion of the dataset used to evaluate the model’s performance and generalization ability after it has been trained.</p><h3>How to Divide Data</h3><p>The most common way to divide data is a simple percentage split. The ideal ratio depends on the size of your dataset, but some common splits are:</p><ul><li><strong>80/20 Split:</strong> 80% of the data is used for training, and 20% is used for testing. This is a very common and effective split for many datasets.</li><li><strong>70/30 Split:</strong> 70% for training and 30% for testing. This is another popular choice, providing a slightly larger test set for more robust evaluation.</li><li><strong>60/40 Split:</strong> 60% for training and 40% for testing. This might be used when you have a very large dataset and can afford to use more data for testing.</li></ul><p>The division should be done <strong>randomly</strong> to ensure that both the training and testing sets are representative of the overall dataset and to avoid any bias in the distribution of data. Most machine learning libraries, like scikit-learn in Python, provide functions to easily split data in this way.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/344/1*SisQCEbO45a7FMJq4l4AXw.png" /></figure><h3><strong>3. What are input features and target variable?</strong></h3><p>In machine learning, <strong>input features</strong> are the independent variables (often denoted as ‘X’) that you feed into a model to make a prediction. The <strong>target variable</strong> is the dependent variable (often denoted as ‘y’) that you are trying to predict.</p><ul><li><strong>Input Features (X):</strong> These are the predictors, attributes, or measurements from your data that are used to predict the outcome. For example, if you want to predict a house’s price, the input features could be its size in square feet, the number of bedrooms, and its location.</li><li><strong>Target Variable (y):</strong> This is the outcome or the value you want the model to learn to predict. In the house price example, the target variable would be the actual price of the house.</li></ul><h3><strong>4. What are parameters?</strong></h3><p>In machine learning, <strong>parameters</strong> are the internal variables of a model that are learned directly from the training data. These are the values that the model “tunes” during the training process to make accurate predictions.</p><p>A classic example of parameters can be found in a linear regression model (y= mx+c). Here, the slope (<strong>m</strong>) and the y-intercept ( c ) are the parameters. The model learns the optimal values for <strong>m</strong> and <strong>c</strong> from the training data to best fit a line to the data points.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/311/1*47MxoNW-CHyZ51CuWPFQVA.png" /></figure><h3><strong>5. What is loss function and cost function?</strong></h3><p>A <strong>loss function</strong>, also known as an error function, quantifies the penalty or “loss” for a single training example. It takes the predicted output and the true target value as inputs and returns a real number representing the error for that specific instance.</p><p>For a single training example with features xi​, a true label yi​, and a predicted label y^​i​, the loss is given by:</p><p>Loss=L(y(i)​,y^​i​)</p><p>A <strong>cost function</strong>, often referred to as an objective function, provides a measure of the overall performance of the model on the entire training dataset. It is typically calculated by averaging the loss over all the training examples.</p><p>For a training set with N examples, the cost function J is:</p><p>Cost=J=N1​i=1∑N​L(yi​,y^​i​)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/274/1*sc7CCbKY7_yvWbrAMq5STQ.png" /></figure><h3><strong>6. What is gradient descent? How to apply it?</strong></h3><p>Gradient descent is an optimization algorithm used to find the minimum value of a function. In machine learning, it’s used to minimize the model’s <strong>cost function</strong> by iteratively adjusting the model’s parameters, like weights and biases.</p><p>Think of it like being lost on a mountain in thick fog and wanting to get to the lowest valley. You can’t see the whole landscape, but you can feel the slope of the ground right where you are. To get down, you’d take a step in the steepest downhill direction. You’d repeat this process, and eventually, you’d reach the bottom. 🏔️</p><p>That’s exactly what gradient descent does. The “slope of the ground” is the <strong>gradient</strong> of the cost function, and each “step” is an update to the model’s parameters</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/308/1*hHJemPoPg5qSD48YB7W2PA.png" /></figure><h3><strong>7. How to apply linear regression with multiple features?</strong></h3><p>While simple linear regression uses the formula y=mx+b, multiple linear regression extends this to include multiple features (x1​,x2​,…,xn​). The formula for the model’s prediction is:</p><p>y=b0​+b1​x1​+b2​x2​+⋯+bn​xn​</p><p>Let’s break this down:</p><ul><li><strong>y</strong> is the dependent variable — the single value you are trying to predict (e.g., house price).</li><li><strong>x1​,x2​,…,xn​</strong> are the independent variables or features (e.g., square footage, number of bedrooms, age of the house).</li><li><strong>b0​</strong> is the <strong>intercept</strong> (or bias) of the model. It’s the predicted value of y if all features were zero.</li><li><strong>b1​,b2​,…,bn​</strong> are the <strong>coefficients</strong> (or weights) for each feature. Each coefficient bi​ represents the change in y for a one-unit change in the corresponding feature xi​, assuming all other features remain constant.</li></ul><p>The main goal is to find the optimal values for the intercept (b0​) and all the coefficients (b1​,…,bn​) during the training process.</p><p>Here is a practical guide to applying multiple linear regression.</p><h4>1. Prepare Your Data</h4><p>Before building the model, your data needs to be ready.</p><ul><li><strong>Gather Data</strong>: Collect your data, ensuring you have multiple feature columns (the xi​ values) and one target column (the y value).</li><li><strong>Feature Scaling</strong>: This step is crucial. When features are on vastly different scales (e.g., square footage from 1,000–5,000 and number of bedrooms from 2–5), the model can have trouble learning. <strong>Scale your features</strong> so they have a comparable range. Common methods include:</li><li><strong>Normalization</strong>: Scales data to a range between 0 and 1.</li><li><strong>Standardization</strong>: Scales data to have a mean of 0 and a standard deviation of 1.</li></ul><h4>2. Formulate the Model</h4><p>Define the model structure based on your features. If you have three features (e.g., size, bedrooms, age), your model’s hypothesis will be:</p><p>price=b0​+b1​(size)+b2​(bedrooms)+b3​(age)</p><h4>3. Find the Optimal Parameters (Model Training) ⚙️</h4><p>The core of the process is finding the best coefficients (bi​) that minimize the model’s error, which is typically measured by the <strong>Mean Squared Error (MSE)</strong> cost function. There are two primary methods for this:</p><ul><li><strong>Gradient Descent</strong>: As discussed previously, this is an iterative approach. It starts with random coefficients and repeatedly adjusts them in the direction that minimizes the MSE until it finds the best possible fit. This is the most common method for very large datasets.</li><li><strong>The Normal Equation</strong>: This is a direct, analytical approach to solving for the optimal coefficients. It computes them in a single step using matrix algebra:</li><li>θ=(XTX)−1XTy</li><li><strong>Pros</strong>: No need to choose a learning rate or iterate. It’s fast for a smaller number of features.</li><li><strong>Cons</strong>: It becomes very slow if you have a huge number of features (e.g., &gt;10,000), as calculating the inverse of a large matrix is computationally expensive.</li></ul><h4>4. Make Predictions</h4><p>Once the model is trained and you have your optimal coefficients, you can use the formula to make predictions on new, unseen data. Simply plug the feature values of the new data point into the equation.</p><h3><strong>8. What is vectarization?</strong></h3><p>Vectorization is the process of converting an algorithm from operating on a single value at a time to operating on a set of values — like a vector or array — all at once. In practice, it means replacing explicit for loops in your code with high-level, optimized library calls.</p><h3>Why is Vectorization Important?</h3><p>The primary benefit of vectorization is <strong>speed</strong>. Instead of performing an operation on one element at a time in a loop, a vectorized operation performs the calculation on an entire array simultaneously. This is significantly more efficient for two main reasons:</p><ol><li><strong>Code Interpretation</strong>: In languages like Python, for loops are interpreted one line at a time, which carries a lot of overhead. Vectorized operations are typically written in low-level, compiled languages (like C or Fortran), making them much faster.</li><li><strong>Hardware Optimization</strong>: Modern CPUs and GPUs are designed for parallel operations. They have special instructions (like SIMD — Single Instruction, Multiple Data) that can perform the same operation on multiple data points at the same time. Vectorized code takes full advantage of this hardware capability.</li></ol><pre>If you liked this post, please follow me for more such high quality posts and do engage with us.</pre><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=aacf994c4d88" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Indian Students Can Now Use Gemini 2.5 Pro and Perplexity Pro for Free — Here’s How]]></title>
            <link>https://medium.com/@sc.gondi2005/indian-students-can-now-use-gemini-2-5-pro-and-perplexity-pro-for-free-heres-how-27992d7b868a?source=rss-485c1dfb243a------2</link>
            <guid isPermaLink="false">https://medium.com/p/27992d7b868a</guid>
            <category><![CDATA[students]]></category>
            <category><![CDATA[gemini]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[google]]></category>
            <category><![CDATA[perplexity-ai]]></category>
            <dc:creator><![CDATA[Suchendra Charith Gondi]]></dc:creator>
            <pubDate>Thu, 17 Jul 2025 15:23:32 GMT</pubDate>
            <atom:updated>2025-07-17T15:23:32.749Z</atom:updated>
            <content:encoded><![CDATA[<h3>Indian Students Can Now Use Gemini 2.5 Pro and Perplexity Pro for Free — Here’s How</h3><blockquote>Hey, if you don’t know about the current <em>offers</em> provided by <strong>Google and Perplexity</strong> don’t worry let me explain you clearly. So, on July 15th Google released an offer stating that from now on Indian Students who are above 18 years with valid college email ids can claim a 1 year Gemini 2.5 Pro version ( costs ₹19,500/year ) . The benefits included in this plan are Deep Research, <strong>Gemini Live, Gemini Pro across all Google apps, NoteBookLM, Veo3</strong> and a whopping <strong>2TB storage</strong> absolutely for free !!!</blockquote><p>If you are wondering how to claim this offer no worries. Let me help you there clearly! So, do follow these steps properly:<br><strong>Step-1 :</strong> Click on this link : <a href="https://gemini.google/students/?gl=IN">https://gemini.google/students/?gl=IN</a><br><strong>Step-2 :</strong> Click on <strong><em>Get Offer</em></strong> Button.<br><strong>Step-3 :</strong> Then it will be redirected to a verification page. Click on <strong>Verify Eligibility</strong> button and then enter the personal details.<br>Step-4 : Also, include the <strong>professional email</strong> with it, then boom you’re officially a Gemini Pro member.</p><blockquote><strong>NOTE:</strong> This offer is available from <strong>15/07/2025 to 15/09/2025</strong> claim it before the offer ends. The <strong>requirements</strong> are very simple — You need to be an <strong>Indian</strong>, you must be<strong> 18+ years old</strong> and you must have a <strong>college email id</strong>.</blockquote><p>Also, Perplexity is offering it’s Pro version anually ( usually worth ₹17,000/year) for free. So, Bharti Airtel — India’s #2 telecom company— announced a <strong>partnership with Perplexity AI</strong> to give <em>all</em> <strong>Airtel customers</strong> a <strong>free 1-year Perplexity Pro subscription. </strong>This covers ~360 million mobile, broadband and DTH subscribers.</p><blockquote>Now, if you wanted to claim this <strong>subscription</strong> , you need to follow these steps:<br><strong>Step-1 :</strong> Go to Airtel Thanks App.<br><strong>Step-2 :</strong> Open the OTT and Rewards Section.<br><strong>Step-3 :</strong> Click on Claim Perplexity Pro gift offer.<br><strong>Step-4 :</strong> Enter the code which will be emailed to your email account.<br>…. and boom enjoy your <strong>free Perplexity Pro</strong> for <strong>1 year</strong> !!!</blockquote><p>If you found out that the blog helped you in any of the form , please consider following me, follow back will be there. Please engage with the blog, criticize, correct me , share your thoughts about this bold step!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=27992d7b868a" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>