<?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 Mete Oğuz on Medium]]></title>
        <description><![CDATA[Stories by Mete Oğuz on Medium]]></description>
        <link>https://medium.com/@meteoguz1308?source=rss-0b8b009fa19b------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*vcQ2z5aZUQFB8nSz2szWcg.jpeg</url>
            <title>Stories by Mete Oğuz on Medium</title>
            <link>https://medium.com/@meteoguz1308?source=rss-0b8b009fa19b------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 15:41:51 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@meteoguz1308/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[How to Explore Data Like a Pro: A Simple Guide]]></title>
            <link>https://medium.com/@meteoguz1308/how-to-explore-data-like-a-pro-a-simple-guide-a4a1419bd474?source=rss-0b8b009fa19b------2</link>
            <guid isPermaLink="false">https://medium.com/p/a4a1419bd474</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[eda]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[data-visualization]]></category>
            <dc:creator><![CDATA[Mete Oğuz]]></dc:creator>
            <pubDate>Tue, 28 Jan 2025 13:51:49 GMT</pubDate>
            <atom:updated>2025-01-28T13:51:49.829Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*K-UjaOEhtHFt3z-RbKMZ_w.jpeg" /></figure><p>Exploratory Data Analysis (EDA) is like getting to know someone before you start working together. It’s about understanding what the data looks like, spotting patterns, and finding problems. Here’s a breakdown of the steps in this EDA process, explained in the simplest way possible.</p><h4>1. Setting Up and Getting Ready</h4><p>First, we import the libraries and make a few settings to ensure everything works smoothly.</p><pre>import numpy as np<br>import pandas as pd<br>import seaborn as sns<br>import matplotlib.pyplot as plt<br>import matplotlib<br>import plotly.express as px</pre><p><strong>What does this do?</strong></p><ul><li>We’re bringing in tools like pandas and numpy to work with data and seaborn and matplotlib to create graphs.</li><li>The set_option lines make sure we can see all rows and columns when we print something. This is super helpful when working with large datasets.</li></ul><h4>2. Loading and Taking a Quick Look at the Data</h4><p>Next, we take a look at its structure.</p><pre>df.head()<br>df.tail()<br>df.shape<br>df.info()<br>df.columns<br>df.index<br>df.describe().T<br>df.isnull().values.any()<br>df.isnull().sum()</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/563/1*0t5g_vL2bF44Mf30TCG1OA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/903/1*N4wNeNagEE7lOxs5MnAnyg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/497/1*gY3HeOeGuqJq7fcsAnQEdw.png" /></figure><p><strong>What does this do?</strong></p><ul><li>df.head() and df.tail() show the first and last few rows of the data. It’s like peeking at the start and end of a book to see what’s inside.</li><li>df.shape tells us how many rows and columns the dataset has.</li><li>df.info() lists the types of data in each column and how many missing values there are.</li><li>df.describe() gives us basic stats like the average and maximum values for each numeric column.</li><li>Checking for missing values (df.isnull().sum()) shows where we might have gaps to fill.</li></ul><h4>3. A Handy Function to Summarize Everything</h4><p>We created a custom function to summarize all this info in one go.</p><pre>def check_df(dataframe, head=5):<br>    print(&quot;##################### Shape #####################&quot;)<br>    print(dataframe.shape)<br>    print(&quot;##################### Types #####################&quot;)<br>    print(dataframe.dtypes)<br>    print(&quot;##################### Head #####################&quot;)<br>    print(dataframe.head(head))<br>    print(&quot;##################### Tail #####################&quot;)<br>    print(dataframe.tail(head))<br>    print(&quot;##################### NA #####################&quot;)<br>    print(dataframe.isnull().sum())<br>    print(&quot;##################### Quantiles #####################&quot;)<br>    print(dataframe.describe([0, 0.05, 0.50, 0.95, 0.99, 1]).T)</pre><pre>check_df(df)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/903/1*N4wNeNagEE7lOxs5MnAnyg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/838/1*bJna8NKOaA9es3R5b0dKLg.png" /></figure><p><strong>What does this do?</strong></p><ul><li>This function combines everything from Step 2 into one simple command. It’s a great way to quickly understand what’s in the data.</li></ul><h4>4. Categorizing Columns</h4><p>Here, we divide the columns into groups like “categorical” and “numerical” to make our analysis easier.</p><pre>def grab_col_names(dataframe, cat_th=10, car_th=20):<br>    cat_cols = [col for col in df.columns if str(df[col].dtypes) in [&quot;category&quot;, &quot;object&quot;, &quot;bool&quot;]]<br>    num_but_cat = [col for col in df.columns if df[col].nunique() &lt; cat_th and df[col].dtypes in [&quot;int64&quot;, &quot;float64&quot;]]<br>    cat_but_car = [col for col in df.columns if df[col].nunique() &gt; car_th and str(df[col].dtypes) in [&quot;category&quot;, &quot;object&quot;]]<br>    cat_cols = cat_cols + num_but_cat<br>    cat_cols = [col for col in cat_cols if col not in cat_but_car]<br>    num_cols = [col for col in df.columns if df[col].dtypes in [&quot;int64&quot;, &quot;float64&quot;]]<br>    num_cols = [col for col in num_cols if col not in cat_cols]<br>    print(f&quot;Observations: {dataframe.shape[0]}&quot;)<br>    print(f&quot;Variables: {dataframe.shape[1]}&quot;)<br>    print(f&#39;cat_cols: {len(cat_cols)}&#39;)<br>    print(f&#39;num_cols: {len(num_cols)}&#39;)<br>    print(f&#39;cat_but_car: {len(cat_but_car)}&#39;)<br>    print(f&#39;num_but_cat: {len(num_but_cat)}&#39;)<br>    return cat_cols, num_cols, cat_but_car, num_but_cat</pre><pre>cat_cols, num_cols, cat_but_car, num_but_cat = grab_col_names(df, cat_th = 10, car_th=20)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/268/1*HtwyPuJuUzkPBs8TLviH3g.png" /></figure><p><strong>What does this do?</strong></p><p>Groups columns into:</p><p><strong>Categorical Columns (</strong><strong>cat_cols)</strong>:</p><ul><li>These include columns with text or categories (like “male” and “female”) or bool values (True/False).</li><li>Columns with numbers but very few unique values (like 1, 2, 3) are also included here.</li></ul><p><strong>Numerical Columns (</strong><strong>num_cols)</strong>:</p><ul><li>These are columns with continuous numbers (like prices, ages, or measurements).</li><li>Columns already classified as categorical are excluded.</li></ul><p><strong>Numerical but Categorical (</strong><strong>num_but_cat)</strong>:</p><ul><li>These are columns that are technically numerical (like integers) but have only a small number of unique values.</li><li><strong>Example:</strong> A column with levels like 1 = Beginner, 2 = Intermediate, 3 = Advanced.</li><li>These are treated as categories because their values represent groups, not continuous measurements.</li></ul><p><strong>Categorical but Cardinal (</strong><strong>cat_but_car)</strong>:</p><ul><li>These are columns with text categories but have too many unique values to be useful in analysis.</li><li><strong>Example:</strong> An ID column where every row has a unique value.</li></ul><p>By grouping the columns, we can apply the right analysis for each type of data. For example, we’ll count and plot categories for cat_cols, but calculate averages and correlations for num_cols. This makes the analysis more focused and efficient!</p><h4>5. Analyzing Categorical Variables</h4><p>This step helps us understand the distribution of categorical variables.</p><pre>def cat_summary(dataframe, col_name, plot=False):<br>    print(pd.DataFrame({col_name: dataframe[col_name].value_counts(),<br>                        &quot;Ratio&quot;: 100 * dataframe[col_name].value_counts() / len(dataframe)}))<br>    print(&quot;##########################################&quot;)<br>    if plot:<br>        sns.countplot(x=dataframe[col_name], data=dataframe)<br>        plt.show()</pre><pre>for col in cat_cols:<br>    cat_summary(df, col, plot=True)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/635/1*yfr45MXTlpUcUAm5SwtKWQ.png" /></figure><p><strong>What It Does</strong>:</p><p>For each categorical column:</p><ul><li>It calculates the <strong>frequency</strong> of each category (how many times each category appears).</li><li>It calculates the <strong>percentage</strong> of the total for each category (how much of the dataset belongs to that category).</li></ul><p><strong>What It Shows</strong>:</p><ul><li><strong>Category Importance</strong>: You can see which categories are most common and which are rare. For example, in a gender column, you might find 60% are male and 40% are female.</li><li><strong>Data Balance</strong>: It helps you check if the data is balanced or skewed. For example, if one category dominates, it might require special handling during analysis.</li></ul><p><strong>Why It’s Useful</strong>:</p><ul><li>It provides a quick and clear understanding of how categorical data is structured.</li><li>It helps in deciding if certain categories need grouping (e.g., merging rare categories) or if adjustments are needed to ensure fair analysis.</li></ul><h4>6. Analyzing Numerical Variables</h4><p>Now, we look at the spread and shape of numerical data.</p><pre>def num_summary(dataframe, numerical_col, plot=False):<br>    quantiles = [0.05, 0.10, 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90, 0.95, 0.99]<br>    print(dataframe[numerical_col].describe(quantiles).T)<br>    if plot:<br>        dataframe[numerical_col].hist(bins=50)<br>        plt.xlabel(numerical_col)<br>        plt.title(numerical_col)<br>        plt.show(block=True)<br>    print(&quot;#####################################&quot;)</pre><pre>for col in num_cols:<br>    num_summary(df, col, True)</pre><pre>df_num = df.select_dtypes(include = [&#39;float64&#39;, &#39;int64&#39;])<br>df_num.hist(figsize=(16, 20), bins=50, xlabelsize=8, ylabelsize=8)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/498/1*UxtskcGibT_4AfZ-5l2vag.png" /></figure><h4>What Does This Step Do?</h4><p>Calculating Statistics:</p><p>For each numerical column, the function calculates key statistics like:</p><ul><li><strong>Mean:</strong> The average value.</li><li><strong>Min/Max:</strong> The smallest and largest values.</li><li><strong>Quantiles:</strong> Values at specific percentiles (e.g., 25th, 50th, 95th percentiles).</li><li>This helps identify how the data is spread out and whether there are extreme values (outliers).</li></ul><p><strong>Visualizing the Distribution:</strong></p><ul><li>The function creates a <strong>histogram</strong> for each numerical column. Histograms show how frequently values appear in specific ranges.</li><li>For example, if you’re analyzing house prices, the histogram might show that most houses fall within a certain price range, with a few very expensive ones.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/615/1*q_TpR1pVD8aKTmefczgQMw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qKISXGKNrhgqRCmm5mvi0Q.png" /></figure><h4>What Does It Show?</h4><ul><li><strong>Data Spread:</strong> Are the values tightly clustered or widely spread out? This can indicate variability in the data.</li><li><strong>Outliers:</strong> Histograms and quantiles can help spot extreme values that might need special attention.</li><li><strong>Skewness:</strong> If most of the values are on one side of the histogram, it shows the data is skewed (e.g., many small values and a few very large ones).</li><li><strong>Central Tendency:</strong> The mean and median tell us where the “center” of the data is, which is helpful for summarizing large datasets.</li></ul><h4>Why Is It Useful?</h4><ul><li>These insights help you decide if the data needs transformation (e.g., log transformation for skewed data).</li><li>Spotting outliers early can help avoid errors in later steps like modeling or predictions.</li><li>It gives you a clear picture of the overall behavior of numerical columns, which can guide decision-making.</li></ul><h4>7. Correlation Analysis</h4><p>We check how numerical variables relate to each other.</p><pre>f, ax = plt.subplots(figsize=[30, 30])<br>sns.heatmap(df[num_cols].corr(), annot=True, fmt=&quot;.2f&quot;, ax=ax, cmap=&quot;YlGnBu&quot;, linewidths=5)<br>ax.set_title(&quot;Correlation Matrix&quot;, fontsize=20)<br>plt.show()</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*17Npn1cVU_OekFsh2NxIBw.png" /></figure><h4>What Does This Step Do?</h4><p><strong>Calculate Correlations:</strong></p><ul><li>The correlation matrix measures how strongly numerical variables are related to each other.</li><li>Correlation values range from <strong>-1 to 1</strong>:</li><li><strong>1</strong> means a perfect positive relationship (as one variable increases, the other also increases).</li><li><strong>-1</strong> means a perfect negative relationship (as one variable increases, the other decreases).</li><li><strong>0</strong> means no relationship between the variables.</li><li>These correlations are calculated using the df[num_cols].corr() function.</li></ul><p><strong>Visualize with a Heatmap:</strong></p><ul><li>A <strong>heatmap</strong> is created using sns.heatmap, which displays the correlation values in a grid.</li><li>Colors indicate the strength of the correlation</li><li>Darker or brighter colors usually mean stronger relationships.</li><li>Lighter or neutral colors mean weaker or no relationships.</li><li>Correlation values are also printed on the heatmap for clarity.</li></ul><h4>What Does It Show?</h4><p><strong>Strong Relationships:</strong></p><ul><li>You can quickly spot variables that are highly correlated. For example, house size might have a strong positive correlation with price.</li></ul><p><strong>Potential Multicollinearity:</strong></p><ul><li>If two or more variables are highly correlated (close to 1 or -1), it could indicate multicollinearity, which might cause issues in machine learning models.</li></ul><p><strong>Unrelated Variables:</strong></p><ul><li>Variables with low or near-zero correlation values don’t have a strong relationship, meaning changes in one variable don’t affect the other.</li></ul><h4>Why Is It Useful?</h4><p><strong>Feature Selection:</strong></p><ul><li>By identifying highly correlated variables, you can decide which ones to keep or remove. For example, if two variables are almost identical, you might only keep one.</li></ul><p><strong>Better Insights:</strong></p><ul><li>It helps you understand which variables influence each other. For instance, if house size strongly correlates with price, it suggests that larger houses are generally more expensive.</li></ul><p><strong>Prepare for Modeling:</strong></p><ul><li>Detecting multicollinearity allows you to take corrective actions, like using dimensionality reduction techniques or dropping redundant features.</li></ul><h4>8. Finding Missing Values</h4><p>Finally, we look at which columns have missing values and how much.</p><pre>def missing_values_table(dataframe, na_name=False):<br>    na_columns = [col for col in dataframe.columns if dataframe[col].isnull().sum() &gt; 0]<br>    n_miss = dataframe[na_columns].isnull().sum().sort_values(ascending=False)<br>    ratio = (dataframe[na_columns].isnull().sum() / dataframe.shape[0] * 100).sort_values(ascending=False)<br>    missing_df = pd.concat([n_miss, np.round(ratio, 2)], axis=1, keys=[&#39;n_miss&#39;, &#39;ratio&#39;])<br>    print(missing_df, end=&quot;\n&quot;)<br>    if na_name:<br>        return na_columns</pre><pre>missing_values_table(df)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/388/1*oMRAI4ShSdRH41-fGufyqg.png" /></figure><p><strong>What does this do?</strong></p><ul><li>Lists the columns with missing values and shows the number and percentage of missing entries.</li></ul><h3>Conclusion</h3><p>Exploratory Data Analysis (EDA) is the foundation of any data-driven project. By following a step-by-step approach, we can gain a thorough understanding of our dataset and prepare it for deeper analysis or modeling. Here’s a summary of what we achieved:</p><ul><li><strong>Data Familiarization:</strong><br>We started by exploring the dataset’s structure, size, and content. This helped us quickly understand what kind of data we’re working with.</li><li><strong>Column Categorization:</strong><br>By grouping columns into categories like categorical, numerical, and high-cardinality, we ensured that each type of data was analyzed appropriately. This categorization simplifies analysis and avoids errors.</li><li><strong>Categorical Variable Insights:</strong><br>We analyzed how often categories appear and their distribution in the dataset. This step highlighted any imbalances or dominant categories, which could affect future analysis or modeling.</li><li><strong>Numerical Variable Distribution:</strong><br>We calculated key statistics and visualized the distribution of numerical data. This step helped identify patterns, outliers, and potential data transformations needed to improve data quality.</li><li><strong>Correlation Analysis:</strong><br>Using a heatmap, we examined relationships between numerical variables. This helped spot strong correlations, which might indicate redundant information or key interactions.</li><li><strong>Handling Missing Values:</strong><br>We identified columns with missing data and calculated their percentages. This step is crucial for deciding how to handle missing values (e.g., filling, removing).</li></ul><h4>Why This Matters:</h4><p>This EDA process provides a solid understanding of the dataset. It ensures that we:</p><ul><li>Uncover hidden patterns.</li><li>Address data issues early (like missing values or outliers).</li><li>Lay the groundwork for building reliable models.</li></ul><p>By mastering EDA, we make smarter, data-informed decisions and build a strong foundation for success in any data project.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a4a1419bd474" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Thriving in the AI and Data-Driven Era]]></title>
            <link>https://medium.com/@meteoguz1308/thriving-in-the-ai-and-data-driven-era-fd75d395f118?source=rss-0b8b009fa19b------2</link>
            <guid isPermaLink="false">https://medium.com/p/fd75d395f118</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[bussiness]]></category>
            <category><![CDATA[future]]></category>
            <category><![CDATA[data-analysis]]></category>
            <dc:creator><![CDATA[Mete Oğuz]]></dc:creator>
            <pubDate>Sun, 19 Jan 2025 21:00:35 GMT</pubDate>
            <atom:updated>2025-01-19T21:00:35.980Z</atom:updated>
            <content:encoded><![CDATA[<p>Artificial intelligence (AI) and data science have become some of the most significant technological advancements in recent years, profoundly transforming the business world and our daily lives. Many people believe that AI will completely take over human labor in the future. However, the advancement of technology does not imply the complete elimination of the human factor. On the contrary, individuals who can adapt to AI, understand its working principles, and guide it efficiently will always be sought-after professionals in the business world.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/950/1*kiPjC9fxFg_EREQDwtww7A.jpeg" /></figure><p>The advancement of AI, the increase in automation, and the digitalization of business processes have led to fundamental changes in many sectors. AI applications are used in various fields, from customer service to manufacturing, and from the healthcare sector to the financial world. For example, AI-supported diagnostic systems in the healthcare sector enable doctors to make faster and more accurate decisions, while fraud detection systems in the financial sector help prevent financial losses.</p><p>However, the successful implementation of these technologies still relies on the human factor playing a crucial role. Individuals who can understand AI’s working principles and guide it effectively can extract meaningful insights from the vast datasets AI provides, thereby making business decision-making processes more efficient. A person capable of working with one of the most innovative and impressive AI models, such as Chat GPT, by writing the right prompts and directing it effectively to achieve the desired outputs, can make a significant difference.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/826/1*QwTIIRA9XsWUmY68Etzk4w.jpeg" /></figure><p>At this point, data science comes into play by analyzing large datasets and providing meaningful insights to businesses. Making decisions based on accurate data is a critical factor in gaining a competitive advantage for enterprises. For example, companies in the e-commerce sector that analyze customer behaviors can personalize their marketing strategies and increase their revenues. Data analysts and data scientists play a vital role in ensuring that AI systems function effectively.</p><p>However, it is important to emphasize that data science and AI are not limited to becoming a professional data scientist. The ability to make decisions based on data and data literacy provides a significant advantage in every profession. Understanding AI and having a data-driven mindset help individuals make better strategic decisions in both business and daily life. Therefore, improving data literacy will enable individuals to act more consciously and data-driven in the future. As interest in data science continues to grow, great opportunities arise for those looking to pursue a career in this field. For instance, data analysis in the healthcare sector can assist in early diagnosis of diseases, while in the finance sector, it can optimize investment strategies across various areas.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Rr_O_7BlCQgu4aRzeUy6Ow.jpeg" /></figure><p>To fully benefit from these technologies, individuals need not only to acquire technical knowledge but also to develop critical thinking skills. Knowing how to use data effectively and understanding which data is important are crucial competencies. Working with AI and data science also requires acting in compliance with ethical standards and utilizing accurate data sources, which is a significant responsibility.</p><p>In conclusion, artificial intelligence and data science are among the most important elements shaping the future of the business world. However, the notion that these technologies will completely replace human labor is incorrect. Individuals who can effectively utilize AI, understand its working principles, and make data-driven decisions will have an indispensable position in the future business world. Being open to continuous learning and development is the key to success in the evolving job market. Instead of perceiving technology as a threat, leveraging it efficiently to create new opportunities should be an important goal for everyone.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ty7EatFqjoYbFCX1iwJ95A.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fd75d395f118" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[“Chatting with SQL: Your Data Dialogue Guide]]></title>
            <link>https://medium.com/@meteoguz1308/chatting-with-sql-your-data-dialogue-guide-a00436ce64d5?source=rss-0b8b009fa19b------2</link>
            <guid isPermaLink="false">https://medium.com/p/a00436ce64d5</guid>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[language]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[sql]]></category>
            <category><![CDATA[query]]></category>
            <dc:creator><![CDATA[Mete Oğuz]]></dc:creator>
            <pubDate>Fri, 06 Sep 2024 21:59:00 GMT</pubDate>
            <atom:updated>2024-09-06T21:59:29.140Z</atom:updated>
            <content:encoded><![CDATA[<p>SQL is an extremely popular tool and a “language” used for data analysis. At first glance, it might seem like just software, but as its name suggests — SQL (Structured Query Language) — it’s actually a language. Understanding this point is crucial for learning SQL and developing the right skills. Jumping into SQL without grasping its core as a language can lead to confusion and make the learning process harder.</p><p><strong>A</strong>s mentioned, SQL is a language, and it’s important to remember that every time you use it, you’re essentially having a conversation with the database. The logic behind SQL commands and the flow of queries mirrors a regular conversation. In this essay, I’ll explain the logic of SQL and help you see it as a language you can speak confidently.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1020/1*R7WjrJpB36LJgxvd8W7dtQ.jpeg" /></figure><blockquote><strong>Let’s Start</strong></blockquote><pre>SELECT * <br>FROM table_name</pre><p>This command is like the opening of a conversation — it’s as if you’re saying “hello” to a friend and politely asking for something. In SQL terms, it translates to “Can you select everything from the table?” SQL is a very good friend; if you ask it the right question in the correct way, it will always respond promptly.</p><p>Now, you’ve successfully retrieved a table from a database. You can see the data and begin your analysis to uncover insights. For example, let’s say the next step is to find the average of something.</p><p><strong><em>So, let’s continue our conversation with SQL:</em></strong></p><p>“I want to see the “average” of a specific column and display it as avg_price from the orders table”</p><pre>SELECT <br>AVG(Price) AS avg_price<br>FROM orders</pre><p>Now you have the result because your “communication” with SQL was effective. But what if you want to see the average price for every product in the table? Keep the conversation going!</p><p><strong><em>You might ask:</em></strong></p><p>“I want to see the ‘average prices’ of ‘every product’ from the ‘orders’ table. I also want to ‘see’ the ‘product names’ and ‘group’ the table by product names.”</p><pre>SELECT <br>Product_Name,<br>AVG(Price) AS avg_price<br>FROM orders<br>GROUP BY Product_Name</pre><p>As you can see, it’s quite similar to a regular conversation. Once you understand this pattern, you’ll be able to build more complex “queries.”</p><blockquote><strong>Let’s Try a More Complex Query</strong>:</blockquote><p>Imagine you need to ask SQL the following:</p><p>“I want to see the revenue of the top-selling products in each country. Show me the revenue, product, and country. Also, assign a row number to order them from highest to lowest, and I want to see the best-selling product in every country.”</p><pre>SELECT Country, Product, Revenue <br>FROM (<br>    SELECT Country, [Description],<br>    SUM(Quantity) AS Num_Of_Pieces_Sold, <br>    SUM(Quantity * Price) AS Revenue,<br>    ROW_NUMBER() OVER (PARTITION BY Country ORDER BY SUM(Quantity * Price) DESC) AS RN <br>    FROM retail_II <br>    GROUP BY Country, [Description]<br>     ) A <br>WHERE RN = 1</pre><p>This query may look complex at first glance, but once you grasp SQL’s conversation-like logic, it becomes much clearer.</p><p>The query uses a function called ROW_NUMBER(), which assigns a unique row number to each record within a partition (in this case, by country) based on the sorting of revenue. By filtering for RN = 1, you get the best-selling product in each country.</p><h4><strong><em>Understanding the Technical Details</em></strong></h4><ul><li><strong>SELECT:</strong> Retrieves the data you’re asking for.</li><li><strong>AVG():</strong> Calculates the average of a column.</li><li><strong>GROUP BY:</strong> Groups data by specified columns, so you can perform calculations like averages on grouped data.</li><li><strong>ROW_NUMBER():</strong> Assigns a unique number to each row within a partition, helping you rank or order data.</li></ul><p>Though there are technical details involved, you can always find solutions to common issues through practice. SQL’s error messages are also quite helpful and can guide you in troubleshooting.</p><h4><strong><em>The Key to SQL Success</em></strong><em>:</em></h4><p>If you want to perform meaningful analysis in SQL, the most important step is formulating the right question with an analytical mindset. Once you have the question in mind, asking SQL in its “language” will yield the results you’re looking for.</p><p><strong><em>After you improve your skills in SQL, you should update your resume to reflect not only your technical skills but also your “language” skills.</em></strong></p><h3><strong>Remember</strong>: Start talking with SQL, and it will talk back!</h3><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a00436ce64d5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Business Understanding as a Tool]]></title>
            <link>https://medium.com/@meteoguz1308/business-understanding-as-a-tool-1cbc5df4b1f7?source=rss-0b8b009fa19b------2</link>
            <guid isPermaLink="false">https://medium.com/p/1cbc5df4b1f7</guid>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[crisp-dm]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[data-analytics]]></category>
            <dc:creator><![CDATA[Mete Oğuz]]></dc:creator>
            <pubDate>Thu, 01 Aug 2024 22:05:15 GMT</pubDate>
            <atom:updated>2024-08-01T22:05:15.748Z</atom:updated>
            <content:encoded><![CDATA[<p>Imagine a big data set with an expectation of an insight that should be found from this data set. Many techniques can be used to approach this set. Some might be efficient and make insights more visible but the question is; Can these methods be used without a system of “understanding”? <br>CRISP-DM is one of the fundamental bases in data analytics. However, sometimes fundamental bases can be overlooked or forgotten while some complex practical techniques or methods are in use. The beginning of CRISP-DM is “business understanding” or in other words, aim, goal, target, etc. Setting this understanding is one of the keys to finding insights and gaining a lift but also “not” forgetting it or being aware of it while doing some technical work on a data set is another key.</p><h4>Step by Step</h4><blockquote>Understanding the data</blockquote><p>The process of a data set starts, after deciding the target or drawing the boundaries of the understanding.</p><p>For example, I used “count” functions and I found mean, mod, SD, and CV to understand the data deeply in my last practice. However, while I was using these methods, I repeated the “my understanding” or “my aim” in my mind. Eventually, it showed me whether the standard deviation of that column is needed or not, the median does mean something. It was such a light to examine values below but this was not a permanent light. This light was enlightening as long as I was aware of it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/535/1*SH9PAg1XC2SoH9EFRKndlQ.png" /></figure><p>After getting the big picture, I need to decide rather fill in the missing values or not. Also, I need to decide which values should be filled and with which method because dealing with every missing value might be a waste of time. Again the repeating “understanding” helped me to choose the method. Trying to find missing values with correlation was a good option because “my aim” was to see the effects of advertising methods on sales. So finding the highest correlation with sales was my method.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/668/1*04vF16QUYDPfYhYwLBTITA.png" /></figure><p>The next step was how to approach outliers. Deleting them or capping them were my options and decided to cap them by using upper and lower limits. The reason why use this method is that every outlier was capped among its category and the effect of every value on sales remains pure. So it reduced the variance and made the picture clearer.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/671/1*JH-qnAoWDzgT009Lpjs1Xw.png" /></figure><p>In the end, I became able to calculate the linear regression and find the formula which I needed while seeking insight and I was able to find the coefficients which have effects on sales. I also was able to make a small projection on how to set the adv. budget to get the highest lift in terms of sales.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/601/1*cJuFQeUm0q_7uAACDiBjsg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rs18s7yjRWtIgE9J4jwulw.png" /></figure><p>The “aim” of this text was not to show how I used Excel formulas, made calculations, etc. It was how I used my “understanding” step by step. In every step, I tried to adjust my “understanding” to choose the method that I would use on that step. So, “business understanding” is not a big title that you write at the beginning of the project, then you forget. In every step the aim should be a referred source and with small “sub-adjustments” it becomes useful in every step and it keeps you away from being lost. It is also a kind of proof which shows that mindset and techniques cannot be separated.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1cbc5df4b1f7" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Data Analytics in Formula 1]]></title>
            <link>https://medium.com/@meteoguz1308/data-analytics-in-formula-1-f1675c49a204?source=rss-0b8b009fa19b------2</link>
            <guid isPermaLink="false">https://medium.com/p/f1675c49a204</guid>
            <category><![CDATA[data-analytics]]></category>
            <category><![CDATA[formula-1]]></category>
            <dc:creator><![CDATA[Mete Oğuz]]></dc:creator>
            <pubDate>Sun, 28 Jul 2024 15:55:20 GMT</pubDate>
            <atom:updated>2024-07-28T16:01:45.857Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>DATA-DRIVEN “RACE MAKING”</strong></p><p>F1 is one of the most popular sports in the world and probably the most popular one among all motorsports. One key to understanding the popularity of F1 is the competition. There is great competition at every level and every base in F1. Drivers should win races for their careers and teams should win to earn profit and gain investments. In this rough environment, every small detail is crucial and there is no excuse to miss a detail or miscalculate anything. Data Analytics plays a critical role in examining and analyzing these details.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/367/1*KdVwzrIW8sOJwd7A3qgURg.jpeg" /></figure><p>It can be indicated that there are two different bases of data analytics in F1. One base is to provide live or general data to fans to make them feel they are part of this great competition but this level is mostly about showing some summarized big-picture data in order to attract fans’ attention more and increase marketing. The other base is the usage of data analytics by teams. This is the more complex and critical level because teams set their strategies and plans, <strong><em>during the race</em></strong>, <strong><em>before or after the race</em></strong>, or <strong><em>during the pre-season training</em></strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*L_VYPYRIWJEZuBCh9rbvYw.jpeg" /><figcaption><a href="https://www.mapfre.com/en/insights/innovation/data-analysis-in-formula-1-the-difference-between-victory-and-defeat/">https://www.mapfre.com/en/insights/innovation/data-analysis-in-formula-1-the-difference-between-victory-and-defeat/</a></figcaption></figure><h4><strong><em>During The Race</em></strong></h4><p>Maybe this is one of the toughest areas to use data analytics because it is live and in most cases, there are only seconds to decide what to do. During the race, data teams get data from various resources such as cars, tires, track, weather, lap timings, and from other teams’ cars and drivers as much as they can evaluate. These complex data are used to set a driving strategy, in which lap the driver makes pit-stop, engine mode, or short/long term strategies to outstrip another team to win the race. Detailed usage of data in these areas can be a topic for another article but in the big picture, data analytics is at the heart of every race because “instant insights” are very critical, to set the strategy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*KB2d31Lb3nhY7xfvHs0VDw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RD7aBdWChTb0i3Adgszk0Q.jpeg" /><figcaption><a href="https://brainstation.io/magazine/red-bull-racings-data-driven-brand-dominance">https://brainstation.io/magazine/red-bull-racings-data-driven-brand-dominance</a></figcaption></figure><h4><strong>Before or After The Race</strong></h4><p>During the preparation period of a race, teams analyze data from the last race, last year’s data from the same track, drivers’ lap timings for the same track, drivers’ performances, and car performance in many aspects like engine performance or aerodynamic performance. All of these data are used for catch insight to be used in upcoming races in order to improve performance. The same process is applied after the race to understand every aspect of the last race.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*1uW-ZQmtZgb-JTnrhBU-fg.jpeg" /><figcaption><a href="https://www.autosport.com/f1/news/the-new-technology-with-potential-to-revolutionise-f1s-spectacle/10467376/">https://www.autosport.com/f1/news/the-new-technology-with-potential-to-revolutionise-f1s-spectacle/10467376/</a></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vBsmjx1HSuqgzZeHEM1BBw.png" /><figcaption><a href="https://www.linkedin.com/pulse/telemetry-data-unveiling-holy-grail-formula-1-insights/">https://www.linkedin.com/pulse/telemetry-data-unveiling-holy-grail-formula-1-insights/</a></figcaption></figure><h4><strong><em>During The Pre-Season Training</em></strong></h4><p>Pre-season training and preparations are extremely critical for every team because there is a chance to enter the season with improved car, driving strategy, and performance. For drivers, the process is almost the same in terms of data analytics. However, for teams, there is an opportunity that comes only one time in a year which is producing the next year’s car. Analytic teams evaluate the performance of last year’s cars from different aspects such as fuel consumption, aerodynamics, durability, etc. In light of this data, teams start to build the next season’s car and new parts, of course, machine learning is used to make a predictive analysis of the new car.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/520/1*0DtFtK6RbZ-pywmRVyZyog.jpeg" /></figure><blockquote>How Data Analytics Has Affected Formula 1 Over Time According to “ChatGPT”</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hmL9p_eLxxzIc-V6417yRQ.png" /></figure><ul><li><strong>1950s (Impact Level 1)</strong>: Minimal use of data analytics, mostly mechanical insights and basic calculations.</li><li><strong>1960s (Impact Level 2)</strong>: Slight improvements with more structured data collection but still rudimentary.</li><li><strong>1970s (Impact Level 3)</strong>: Beginning of more systematic data collection and basic analysis.</li><li><strong>1980s (Impact Level 4)</strong>: Introduction of electronic data logging, allowing for better data accuracy and storage.</li><li><strong>1990s (Impact Level 5)</strong>: Telemetry systems become common, enabling real-time data transmission and analysis during races.</li><li><strong>2000s (Impact Level 7)</strong>: Advanced simulation software and predictive analytics tools improve car design, strategy, and performance.</li><li><strong>2010s (Impact Level 9)</strong>: Big data and machine learning significantly enhance decision-making processes, providing deeper insights and better predictions.</li><li><strong>2020s (Impact Level 10)</strong>: AI and cloud computing integration lead to extensive data-driven decision-making, optimizing nearly all aspects of the sport.</li></ul><p>To sum up, Data Analytics is used in F1 at many different levels. Every team gives big importance to data analytics and they use it at every 4 stages of data analytics Descriptive — Diagnostics — Predictive — Prescriptive. It can also be indicated that data analytics is one of the key elements that support the competition and marketing in F1.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F3kNX3jWehQY%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D3kNX3jWehQY&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F3kNX3jWehQY%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/8bc88baa6ab6402c758a396fb9a12545/href">https://medium.com/media/8bc88baa6ab6402c758a396fb9a12545/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f1675c49a204" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>