<?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 Deva Dathan on Medium]]></title>
        <description><![CDATA[Stories by Deva Dathan on Medium]]></description>
        <link>https://medium.com/@dev.d.dathan?source=rss-109355d83d43------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Ir_2kriYaiPJy7uCkJlVFg.png</url>
            <title>Stories by Deva Dathan on Medium</title>
            <link>https://medium.com/@dev.d.dathan?source=rss-109355d83d43------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 09:28:02 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@dev.d.dathan/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[Measure of Association (MOA) in Predictive Modeling — A Beginner’s Guide]]></title>
            <link>https://medium.com/@dev.d.dathan/measure-of-association-moa-in-predictive-modeling-a-beginners-guide-f208254ad0c5?source=rss-109355d83d43------2</link>
            <guid isPermaLink="false">https://medium.com/p/f208254ad0c5</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[data-visualization]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Deva Dathan]]></dc:creator>
            <pubDate>Sun, 03 Aug 2025 07:47:16 GMT</pubDate>
            <atom:updated>2025-08-03T07:47:16.058Z</atom:updated>
            <content:encoded><![CDATA[<p>Measure of Association (MOA) in Predictive Modeling — A Beginner’s Guide</p><p>In predictive modeling, understanding the relationship between variables is a crucial first step. This is where the Measure of Association (MOA) comes into play.</p><p>Today, we’ll cover what MOA is, why it&#39;s important, when to use it, and how it works — all in simple terms.</p><p>---</p><p>🧠 What Is Measure of Association?</p><p>A Measure of Association helps us understand how two variables are related to each other. In statistical terms, it&#39;s a way to quantify the strength and direction of a relationship.</p><p>MOA is commonly referred to as:</p><p>Covariance</p><p>Pearson Correlation Coefficient (a standardized version of covariance)</p><p>Both are used to study how two variables move together.</p><p>---</p><p>📐 Covariance: The First Step</p><p>Covariance is the basic measure that tells us whether two variables move in the same direction.</p><p>Covariance Formula:</p><p>\text{Cov}(X, Y) = \frac{\sum{(X_i - \bar{X})(Y_i - \bar{Y})}}{n - 1}</p><p>&gt; We use n - 1 (instead of just n) to get an unbiased estimate of the population covariance.</p><p>How to Interpret Covariance:</p><p>Positive and large: Variables move together (positive association)</p><p>Negative: One variable increases while the other decreases (negative association)</p><p>Close to 0: No clear relationship</p><p>&gt; Also, if you compute Cov(X, X), you’ll get Var(X) — the variance of X.<br>That’s because the covariance of a variable with itself is just its own variance.</p><p>---</p><p>📊 Variance: A Measure of Spread</p><p>Before we jump into correlation, let&#39;s quickly touch on variance, which measures the spread or dispersion of a single variable.</p><p>\text{Var}(X) = \frac{\sum{(X_i - \bar{X})^2}}{n - 1}</p><p>---</p><p>⚠️ The Problem With Covariance</p><p>One major issue with covariance is that it&#39;s sensitive to the units of measurement.</p><p>For example:</p><p>If height is measured in meters, the covariance will be different than if height is measured in centimeters.</p><p>This makes comparisons tricky.</p><p>---</p><p>✅ Solution: Standardization → Correlation</p><p>To overcome this problem, we normalize the covariance. This leads us to Pearson Correlation Coefficient:</p><p>\text{Corr}(X, Y) = \frac{\text{Cov}(X, Y)}{\sigma_X \cdot \sigma_Y}</p><p>Where:</p><p>= standard deviation of X</p><p>= standard deviation of Y</p><p>This removes the units and keeps the values between -1 and +1.</p><p>&gt; 📌 The units from the numerator (Cov(X, Y)) cancel out with the units in the denominator (standard deviations of X and Y).</p><p>---</p><p>💡 Key Properties of Correlation</p><p>Corr(X, X) = 1<br>A variable is perfectly correlated with itself.</p><p>Range:</p><p>-1 \leq \text{Corr}(X, Y) \leq 1</p><p>+1 → Perfect positive correlation</p><p>-1 → Perfect negative correlation</p><p>0 → No correlation</p><p>Positive correlation lies in the 1st and 3rd quadrants</p><p>Negative correlation lies in the 2nd and 4th quadrants</p><p>&gt; ❗ Note: Both covariance and correlation only capture linear relationships between variables. Non-linear patterns require other techniques.</p><p>---</p><p>📌 When to Use MOA?</p><p>Use Measure of Association when:</p><p>You&#39;re exploring relationships between features in your data.</p><p>You want to select features with the strongest predictive potential.</p><p>You need to check for multicollinearity in regression.</p><p>You&#39;re standardizing variables to remove units or scale differences.</p><p>---</p><p>Absolutely. Here&#39;s a concise version you can include under a new section in your article:</p><p>---</p><p>🧮 Where Can We Use MOA?</p><p>Measure of Association (MOA) — especially correlation and covariance — is widely used across various machine learning and statistical techniques. Here&#39;s where it plays an important role:</p><p>Linear &amp; Logistic Regression:<br>Use correlation to select features with strong predictive power and to avoid multicollinearity.</p><p>PCA (Principal Component Analysis):<br>Built on the covariance matrix to reduce dimensionality while retaining information.</p><p>Time Series Models (like ARIMA):<br>Use autocorrelation to detect patterns over time.</p><p>Neural Networks:<br>MOA helps during preprocessing — to select relevant features and reduce noise.</p><p>Tree-Based Models (Random Forest, XGBoost, etc.):<br>Don’t require MOA directly, but it can still be used for feature selection beforehand.</p><p>Clustering (e.g., K-Means):<br>MOA helps in preprocessing to remove redundant features.</p><p>&gt; In short, MOA is a valuable tool during exploratory data analysis, feature selection, and dimensionality reduction — even if not always part of the algorithm itself.</p><h3>___</h3><p>Absolutely! Here&#39;s a shorter and crisper version of that section, ready to be added under your MOA article:</p><p>---</p><p>🔧 How MOA Helps Build Better Models</p><p>Measure of Association (MOA) isn’t just for understanding relationships — it actively improves model performance. Here’s how:</p><p>1. 🎯 Smarter Feature Selection</p><p>MOA helps identify features that are strongly related to the target.<br>✅ Keep the important ones<br>❌ Drop the irrelevant ones<br>This reduces noise and improves accuracy.</p><p>2. 🚫 Avoiding Multicollinearity</p><p>Highly correlated features can confuse models like linear regression.<br>MOA helps detect and remove such redundancy, making models more stable and interpretable.</p><p>3. 📉 Better Generalization</p><p>By removing weak or duplicate features, the model becomes simpler and performs better on new data (less overfitting).</p><p>4. 🧹 Guides Data Preparation</p><p>MOA highlights when you need to standardize or normalize features — especially when dealing with covariance.</p><p>---</p><p>&gt; In short, MOA helps you clean, select, and structure your data — laying the foundation for better predictive models.</p><p>---</p><p>🔚 Final Thoughts</p><p>Understanding Measure of Association is fundamental in predictive modeling. It helps you decide which variables might influence others and guides you in building better models.</p><p>In the next posts, we’ll dive deeper into how to use correlation in feature selection and how to handle non-linear relationships.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f208254ad0c5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[EDA Made Easy: A Beginner’s Guide to Exploring Data]]></title>
            <link>https://medium.com/@dev.d.dathan/eda-made-easy-a-beginners-guide-to-exploring-data-291b8ad772e1?source=rss-109355d83d43------2</link>
            <guid isPermaLink="false">https://medium.com/p/291b8ad772e1</guid>
            <dc:creator><![CDATA[Deva Dathan]]></dc:creator>
            <pubDate>Sat, 04 Jan 2025 14:05:06 GMT</pubDate>
            <atom:updated>2025-01-04T14:05:06.362Z</atom:updated>
            <content:encoded><![CDATA[<p>Here’s a simple and engaging article on Exploratory Data Analysis (EDA):</p><p><strong>Exploratory Data Analysis (EDA): Understanding Your Data Step by Step</strong></p><p>Before jumping into fancy algorithms or making predictions, there’s a critical step every data analyst takes: <strong>Exploratory Data Analysis (EDA)</strong>. In simple terms, EDA is like meeting your data for the first time and getting to know it. This step ensures you understand its structure, patterns, and quirks before diving deeper.</p><p>Let’s explore what EDA is and how you can approach it, step by step.</p><p>— -</p><p><strong>What is EDA?</strong></p><p>Think of your data as a jigsaw puzzle. You can’t solve it unless you look at the pieces and understand how they fit together. EDA is this process of looking at the “pieces” of your dataset.</p><p>It involves: <br>- Identifying the type of data you have. <br>- Finding patterns, trends, or relationships. <br>- Spotting outliers (unusual values) or missing data.</p><p>— -</p><p><strong>Why is EDA Important?</strong></p><p>Imagine you’re baking a cake. If your ingredients are spoiled or measured incorrectly, the cake will flop. Similarly, EDA helps ensure your “ingredients” (the data) are clean and ready to use. A good EDA can prevent mistakes later on and give you a clearer picture of what’s going on in your dataset.</p><p>— -</p><p><strong>Steps to Perform EDA</strong></p><p>Here’s a simple guide to approaching EDA, with both everyday and technical examples:</p><p><strong>1. Understand Your Data <br></strong>Ask basic questions: <br>- How many rows and columns are there? (Size of the dataset) <br>- What types of data do you have? (Numbers, text, dates, etc.)</p><p><strong>Example:<br></strong>If you’re analyzing sales data, check whether you have information like product names, prices, quantities, and dates.</p><p><strong>Tools:</strong> Use functions like `head()`, `shape`, or `info()` in Python’s pandas library.</p><p>— -</p><p><strong>2. Clean the Data<br></strong>Look for messy data: <br>- Are there missing values? <br>- Are there duplicates? <br>- Are all entries in the right format?</p><p><strong>Example: <br></strong>In a list of phone numbers, some might have country codes while others don’t. Cleaning means ensuring consistency.</p><p><strong>Tools:</strong> Python’s `isnull()` or `drop_duplicates()` can help.</p><p>— -</p><p><strong>3. Summarize the Data<br></strong>Get a quick overview: <br>- What is the average, minimum, or maximum value of a column? <br>- How often does a certain category appear?</p><p><strong>Example:<br></strong>If you’re analyzing exam scores, calculate the average score to understand overall performance.</p><p>Tools: Use `describe()` in Python or Excel’s summary statistics.</p><p>— -</p><p><strong>4. Visualize the Data<br></strong>Create graphs to uncover hidden patterns. <br>- Use bar charts, line graphs, or scatter plots to explore relationships. <br>- Histograms help you understand how data is distributed.</p><p><strong>Example:<br></strong>In sales data, a bar chart can show which product sells the most. A scatter plot can reveal if higher prices lead to fewer sales.</p><p><strong>Tools:</strong> Matplotlib, Seaborn, or Tableau are great for visualization.</p><p>— -</p><p><strong>5. Look for Relationships and Patterns<br></strong>- Are two variables connected? For example, does age affect income? <br>- Do certain groups behave differently?</p><p><strong>Example</strong>:<br>In a customer dataset, check if younger customers tend to shop more online compared to older ones.</p><p><strong>Tools:</strong> Use correlation (`df.corr()`) or group-by operations in Python.</p><p>— -</p><p><strong>Final Thoughts</strong></p><p>EDA is like a detective’s investigation — you’re uncovering the story your data wants to tell. By following these steps, you’ll build a strong foundation for further analysis, whether that means predictive modeling or creating visual reports.</p><p>Remember, the key is curiosity. Ask questions, test ideas, and always double-check your findings. With EDA, you’re setting yourself up for success in any data project.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=291b8ad772e1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Understanding Degrees of Freedom with Everyday Examples]]></title>
            <link>https://medium.com/@dev.d.dathan/understanding-degrees-of-freedom-with-everyday-examples-91c05f9e3a6f?source=rss-109355d83d43------2</link>
            <guid isPermaLink="false">https://medium.com/p/91c05f9e3a6f</guid>
            <dc:creator><![CDATA[Deva Dathan]]></dc:creator>
            <pubDate>Sat, 04 Jan 2025 13:56:44 GMT</pubDate>
            <atom:updated>2025-01-04T13:56:44.501Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Understanding Degrees of Freedom (DOF) in Simple Terms</strong></p><p>Have you ever heard the term “degrees of freedom” and wondered what it really means? Don’t worry — it’s not as complicated as it sounds. In simple terms, <strong>degrees of freedom (DOF)</strong> tell you how many values in a calculation are free to vary while still meeting certain rules or constraints. Let’s break it down.</p><p>— -</p><p><strong>An Everyday Example</strong></p><p>Imagine you’re picking three numbers that must add up to 10. You can freely choose the first two numbers, but the third number is automatically determined by the rule: <br>Sum=10</p><p>For instance: <br>- If you choose 4 and 3, the third number must be 3. <br>- If you choose 6 and 2, the third number must be 2.</p><p>Here, you have <strong>2 degrees of freedom</strong> because the first two numbers are free to vary. The third number isn’t — it’s fixed by the sum.</p><p>— -</p><p><strong>What Does This Mean in Statistics?</strong></p><p>In statistics, <strong>degrees of freedom</strong> refer to the number of independent values in a calculation that can change without breaking any constraints. For example:</p><p>- When calculating the <strong>mean</strong> of nn numbers, once n−1n−1 values are known, the last value is fixed to make the total equal to n×mean. So, there are n−1 degrees of freedom.</p><p>— -</p><p><strong>Why Are Degrees of Freedom Important?</strong></p><p><strong>1. In Variance and Standard Deviation:<br></strong> When calculating sample variance or standard deviation, we divide by n−1 instead of n. Why? Because calculating the mean uses up 1 degree of freedom, leaving n−1 values free to vary. This adjustment ensures our estimates are unbiased.</p><p><strong>2. In Regression Analysis: <br></strong> Degrees of freedom show how many parameters (like predictors) can be freely estimated based on the available data. More predictors = fewer degrees of freedom for error estimation.</p><p>— -</p><p><strong>A Quick Rule of Thumb</strong></p><p>To calculate degrees of freedom, use this simple formula:<br>DOF=Total data points−ConstraintsDOF=Total data points−Constraints<br>Constraints could be the mean, fixed relationships, or model parameters. — -</p><p><strong>Key Takeaway:<br></strong>Degrees of freedom help us understand how much “flexibility” we have in our calculations while following specific rules. Whether it’s a simple sum or a complex statistical model, DOF plays a critical role in ensuring accurate and unbiased results.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=91c05f9e3a6f" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>