<?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 Mariambaydoun on Medium]]></title>
        <description><![CDATA[Stories by Mariambaydoun on Medium]]></description>
        <link>https://medium.com/@mariambaydoun488?source=rss-9178e1ac9302------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*bCmFbEkDQCNPgEN4</url>
            <title>Stories by Mariambaydoun on Medium</title>
            <link>https://medium.com/@mariambaydoun488?source=rss-9178e1ac9302------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 13:55:51 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mariambaydoun488/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[Enhancing Machine Learning Success through Effective Data Collection]]></title>
            <link>https://medium.com/munchy-bytes/enhancing-machine-learning-success-through-effective-data-collection-1f0461fca769?source=rss-9178e1ac9302------2</link>
            <guid isPermaLink="false">https://medium.com/p/1f0461fca769</guid>
            <category><![CDATA[data-processing]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[etl-pipeline]]></category>
            <category><![CDATA[data-collection]]></category>
            <dc:creator><![CDATA[Mariambaydoun]]></dc:creator>
            <pubDate>Tue, 13 Feb 2024 20:34:32 GMT</pubDate>
            <atom:updated>2024-02-13T20:34:32.848Z</atom:updated>
            <content:encoded><![CDATA[<p>Data is crucial for machine-learning models. Our projects would not be possible without it. While you may have worked with organized datasets for AI tasks such as K-Means clustering or logistic regression, real-world data frequently has quality issues or may be unavailable. In the era of big data, precise and scalable data collection methods are critical. Improving our data gathering procedures in effective ways aids the success of machine learning projects.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ETwrBdauzbPCk3I5wI28Rg.png" /><figcaption>Data Collection (<a href="https://www.google.com/url?sa=i&amp;url=https%3A%2F%2Fdatainnovation.org%2F2022%2F11%2Fbetter-data-collection-will-reduce-digital-inequalities%2F&amp;psig=AOvVaw1eBgK6d2oze-FNOM2ISyVU&amp;ust=1707853711081000&amp;source=images&amp;cd=vfe&amp;opi=89978449&amp;ved=0CBQQjhxqFwoTCODQs4PJpoQDFQAAAAAdAAAAABAD">center for data innovation</a>)</figcaption></figure><p>In this article, we discussed the importance of data collecting in the field of machine learning. We looked at several forms of data and highlighted the importance of understanding their characteristics. We also addressed the ETL (Extract, Transform, Load) process, which is critical in gathering and analyzing data for machine learning models.</p><h3><strong>What is Data?</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*s6S4nWVS9Bie8x7uPEA08w.png" /><figcaption>What is Data? (<a href="https://images.app.goo.gl/StyGtHcGgoNVZLKS7">https://www.shiksha.com/</a>)</figcaption></figure><p>Data is commonly defined as a collection of discrete or continuous values that transmit information, such as quantity, quality, fact, statistics, or simply sequences of symbols that may be formally understood.</p><p>Machine Learning relies on data. It refers to the collection of data or statistics that may be used to train a machine learning model. The quality and quantity of data available for training and testing are important factors in determining the accuracy of a machine learning model.</p><h4><strong>Types of Data</strong></h4><p>1.<strong> Structured Data:</strong></p><ul><li><strong>Definition: </strong>Data organized in a tabular format with rows and columns.</li><li><strong>Example: </strong>Databases, spreadsheets.</li><li><strong>Use in ML: </strong>Commonly used in supervised learning, where algorithms learn patterns from labeled datasets.</li></ul><p><strong>2. Unstructured Data:</strong></p><ul><li><strong>Definition:</strong> Data without a predefined data model or structure.</li><li><strong>Example:</strong> Text, images, audio, video.</li><li><strong>Use in ML:</strong> Applied in natural language processing (NLP), computer vision, and audio analysis using techniques like deep learning</li></ul><p>3. <strong>Labeled Data:</strong></p><ul><li><strong>Definition:</strong> Data with predefined labels or categories.</li><li><strong>Example:</strong> Images of cats labeled as “cat” or “dog.”</li><li><strong>Use in ML:</strong> Essential for supervised learning, where the algorithm learns from input-output pairs.</li></ul><p>4. <strong>Unlabeled Data:</strong></p><ul><li><strong>Definition:</strong> Data without predefined labels.</li><li><strong>Example:</strong> Raw text, raw images.</li><li><strong>Use in ML:</strong> Often used in unsupervised learning, where the algorithm discovers patterns and relationships within the data.</li></ul><p>5. <strong>Time Series Data:</strong></p><ul><li><strong>Definition:</strong> Data collected over successive points in time.</li><li><strong>Example:</strong> Stock prices, weather data.</li><li><strong>Use in ML:</strong> Common in forecasting and trend analysis using algorithms designed to understand sequential patterns.</li></ul><p>6. <strong>Spatial Data:</strong></p><ul><li><strong>Definition:</strong> Data related to geographic locations or spatial coordinates.</li><li><strong>Example:</strong> Maps, satellite imagery.</li><li><strong>Use in ML:</strong> Used in applications such as geospatial analysis, urban planning, and environmental monitoring.</li></ul><p>7. <strong>Categorical Data:</strong></p><ul><li><strong>Definition:</strong> Data representing categories or labels.</li><li><strong>Example:</strong> Colors, types of fruits.</li><li><strong>Use in ML:</strong> Requires encoding for machine learning algorithms to process, often used in classification tasks.</li></ul><p>8. <strong>Numerical Data:</strong></p><ul><li><strong>Definition:</strong> Data represented by numbers.</li><li><strong>Example:</strong> Heights, weights.</li><li><strong>Use in ML:</strong> Easily processed by machine learning algorithms, used in regression and clustering.</li></ul><p>In conclusion, understanding the various types of data and their applications in machine learning is essential for developing effective algorithms and models.</p><h4>What is data collection in machine learning?</h4><p>Data collection in machine learning is the process of collecting relevant data that is compatible with your project’s goals and objectives. Finally, this results in the creation of a dataset, which is required for training and feeding into an ML model.</p><h4>Importance of Data Collection</h4><p>Data collection enables the recording of past events, facilitating data analysis to identify recurring patterns. These patterns serve as the foundation for building predictive models with machine learning algorithms, which analyze trends and forecast future changes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*d5oBhJlzS2L1TeolJj8CgA.png" /><figcaption>A Survey On Data Collection for Machine Learning (<a href="https://morioh.com/a/fdb43cf6e7bb/5-essential-papers-on-ai-training-data-kdnuggets">morioh.com</a>)</figcaption></figure><p>Predictive models are only as good as the data from which they are built, so good data collection practices are crucial to developing high-performing models.</p><h4>How much data is needed?</h4><p>When determining the volume of data required for a machine learning model, several factors come into play, including the complexity of the problem, the model, data quality and accuracy, and the availability of labeled data.</p><p>Estimating the amount of data needed can be approached in two ways:</p><ol><li><strong>Rule-of-thumb approach:</strong></li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/572/1*az55PD56H_RDQb4LHy_K_A.png" /><figcaption>Range Rule of Thumb (<a href="https://statisticsbyjim.com/basics/range-rule-of-thumb/">statisticsbyjim.com</a>)</figcaption></figure><p>Smaller datasets are often handled with a rule-of-thumb method. For optimal results in machine learning, your dataset should include at least 10 times as many rows (data points) as features (columns). For example, if your dataset has ten columns (i.e., features), aim for at least 100 rows. However, when working with larger datasets, it is essential to use statistical approaches to accurately estimate the sample size.</p><p><strong>2. Statistical Methods:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*z3Wb-3nNixiLqES8zaHjqA.png" /><figcaption><strong>Statistical Methods (by the author)</strong></figcaption></figure><p>Estimating the required amount of data for statistical analysis involves several methodologies:</p><ul><li>Power analysis assesses the sample size needed for a desired level of statistical power, considering factors like effect size and significance level.</li><li>Margin of error calculations, commonly used in surveys, incorporate confidence level, standard deviation, and desired margin of error.</li><li>Cohen’s effect size provides insight into the magnitude of an effect, helping determine sample size requirements.</li><li>Bootstrap resampling and cross-validation assess variability and model performance in different subsets of data.</li><li>Sequential analysis allows for ongoing analysis, adjusting the sample size as needed.</li><li>Simulation studies, using techniques like Monte Carlo simulations, simulate datasets to evaluate statistical method performance under various conditions.</li></ul><p>These methods collectively aid researchers in determining an appropriate sample size for reliable and meaningful statistical analyses.</p><h3>How is data collected?</h3><p>ETL (Extract, Transform, Load) is the process of collecting data for machine learning models from various sources, transforming it to the desired format, and then loading it to its destination.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*0dZ5rAqAE_cUYL09M7wj_g.png" /></figure><h4>1. Data Extraction (E):</h4><p>This is where the data is gathered from multiple sources, such as surveys, observations, experiments, web scraping, documents, or real-time stock market transactions.</p><p>This data can be raw, and it can be text, numerical values, images, or any other type of information.</p><h4>2. Data Transformation (T):</h4><p>The transformative stage processes the extracted data and prepares it for the intermediate staging environment. It includes cleaning the data to handle missing values, removing duplicates, correcting errors to ensure data quality, and converting data formats (e.g., JSON to database tables). Also, it includes feature engineering, which is creating new features or modifying existing ones to improve the performance of machine learning models. As well as normalization to ensure that different features contribute equally to the model, data integration, and dimensionality reduction through techniques like PCA (Principal Component Analysis),.</p><h4>3. Data Loading (L):</h4><p>This stage focuses on loading the transformed data into the target system. Such as a relational database, a data warehouse, or a distributed file system. It is also important to split the data into training, validation, and test sets. This is crucial for evaluating the machine learning model’s performance.</p><p>Following these stages, the process may proceed to data visualization, documentation, and monitoring, if necessary.</p><h3><strong>Conclusion</strong></h3><p>The article highlights the importance of data collection for successful machine learning programs. It discusses various types of data, shows the importance of understanding them, and outlines the ETL process for effective data handling. The essential conclusion is that the accuracy and success of machine learning models are largely dependent on data quality.</p><h3>References</h3><p>[1] Jackson, E. (2023, September 25). Data ETL Techniques. <em>Level Up</em>. <a href="https://levelup.gitconnected.com/data-etl-techniques-2e3f314f22f3">https://levelup.gitconnected.com/data-etl-techniques-2e3f314f22f3</a></p><p>[2] CrowdForce. (2018, October 23). Data Collection in 4 simple steps. <em>Medium</em>. <a href="https://medium.com/crowdforce-series/data-collection-in-4-simple-steps-5a89044597ef">https://medium.com/crowdforce-series/data-collection-in-4-simple-steps-5a89044597ef</a></p><p>[3] DP6 Team. (2023, March 14). Data Management Series — 2. Data Collection. <em>Medium</em>. <a href="https://medium.com/dp6-us-blog/data-management-series-2-data-collection-178ee48d87c5">https://medium.com/dp6-us-blog/data-management-series-2-data-collection-178ee48d87c5</a></p><p>[4] K, A. (2023, December 19). The Foundation of Insight: Understanding the Importance of Data Collection in Data Analysis. <em>Medium</em>. <a href="https://medium.com/@amulya_k/the-foundation-of-insight-understanding-the-importance-of-data-collection-in-data-analysis-30c1ca36857f">https://medium.com/@amulya_k/the-foundation-of-insight-understanding-the-importance-of-data-collection-in-data-analysis-30c1ca36857f</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1f0461fca769" width="1" height="1" alt=""><hr><p><a href="https://medium.com/munchy-bytes/enhancing-machine-learning-success-through-effective-data-collection-1f0461fca769">Enhancing Machine Learning Success through Effective Data Collection</a> was originally published in <a href="https://medium.com/munchy-bytes">Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>