<?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 Nneoma Ezealuma (dmapgirl) on Medium]]></title>
        <description><![CDATA[Stories by Nneoma Ezealuma (dmapgirl) on Medium]]></description>
        <link>https://medium.com/@dmapgirl?source=rss-bbfcc9f38a59------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*wI567wCR-ZLz2QTZBCQ5tQ.jpeg</url>
            <title>Stories by Nneoma Ezealuma (dmapgirl) on Medium</title>
            <link>https://medium.com/@dmapgirl?source=rss-bbfcc9f38a59------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 11:31:52 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@dmapgirl/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[I Could Have Never Imagined Using Milk and Petroleum to Study Data Analytics]]></title>
            <link>https://dmapgirl.medium.com/i-could-have-never-imagined-using-milk-and-petroleum-to-study-data-analytics-00c15aac6cb8?source=rss-bbfcc9f38a59------2</link>
            <guid isPermaLink="false">https://medium.com/p/00c15aac6cb8</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[data-analysis]]></category>
            <dc:creator><![CDATA[Nneoma Ezealuma (dmapgirl)]]></dc:creator>
            <pubDate>Sun, 01 Nov 2026 02:36:33 GMT</pubDate>
            <atom:updated>2026-11-01T02:44:38.375Z</atom:updated>
            <content:encoded><![CDATA[<h4>But I did, still on my Journey with Python, Pandas, SQL, and Data Visualisation</h4><h4><strong>What Does a Carton of Milk Have to Do with Data Analytics?</strong></h4><p>Imagine walking into a supermarket to buy a carton of milk.</p><p>You pick it up, check the price, and probably don’t think much about where it came from. But behind that simple carton could be an entire story involving farmers, manufacturers, international suppliers, transportation companies, and millions of dollars in trade.</p><p>Now imagine filling up your car at a petrol station. Again, it’s an everyday activity, but behind the fuel pump is another complicated story involving countries, imports, exports, and global trade relationships.</p><p>What do these two everyday products have in common? <strong>Data. Lots of it.</strong></p><p>And that is where my Week 8 data analytics journey began.</p><p>I was so happy I sourced my datasets all by myself last week, and it taught me one thing: collecting data is only the beginning. The real value comes from asking the right questions, organising the information, identifying meaningful patterns, and communicating the findings in a way that others can understand.</p><p>I explored more advanced techniques in Python, Pandas, SQL, and data visualisation. This week was particularly interesting because I moved beyond simply displaying rows of data. I started learning how to transform raw datasets into useful summaries, compare trading activities across countries, and create charts that tell a meaningful story.</p><p>I worked with United Kingdom milk trade records and Nigeria’s 2023 petroleum-product trade data. Both datasets gave me practical opportunities to apply the concepts I was learning.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*78aSuKOg8cTQllAaKzglZw.jpeg" /></figure><h4>1. Understanding Grouping and Aggregation in Pandas</h4><p>One of the most important concepts I learned this week was <strong>Split–Apply–Combine</strong>.</p><p>The idea is straightforward:</p><p>We split a dataset into groups based on selected columns, apply calculations to each group, and combine the results into a summary.</p><p>With Pandas, the groupby() function makes this possible.</p><p>For example:</p><pre>milk.groupby([&#39;Trade Flow&#39;, &#39;Commodity Code&#39;])[&#39;Trade Value (US$)&#39;].sum()</pre><p>This groups milk trade records by trade flow and commodity code, then calculates the total trade value for each combination.</p><p>I also practised different aggregation operations, including sum(), mean(), min(), max(), and aggregate().</p><p>An important lesson was understanding that these functions answer different questions. A total tells us the overall trading value, whereas an average describes typical observations. A maximum identifies the largest individual value, which is not necessarily the month with the highest total.</p><p>I also exploredapply(), which allows custom functions to operate on groups of records.</p><p>This helped me understand that grouping is not simply a coding technique. It is a practical method for answering business questions about performance, activity, and comparisons.</p><h4>2. Exploring the United Kingdom’s Milk Trade</h4><p>The UK’s milk and cream trade dataset formed an important part of the week’s exercises.</p><p>I analysed two milk product classifications:</p><ul><li><strong>HS 0401:</strong> Milk and cream that are neither concentrated nor sweetened.</li><li><strong>HS 0402:</strong> Milk and cream that are concentrated or sweetened.</li></ul><p>The coursework involved both a 2014 monthly dataset and a separate January–May 2015 dataset.</p><p>Using Pandas, I explored trading relationships, import and export activity, leading partner countries, and how trade values changed across reporting periods.</p><p>The 2015 take-home project focused on four important questions: How much did the UK import and export? Who were its major trading partners? Which countries were regular customers? And which countries traded with the UK in both directions?</p><p>These questions made the exercise feel like a real business analysis rather than an isolated programming assignment.</p><p>For example, identifying regular customers required more than simply ranking countries by total trade value. It involved examining whether countries purchased both milk classifications consistently across the five reporting months.</p><p>I also used pivot tables to place imports and exports side by side, making it easier to understand the country’s trade relationships.</p><p>This was a valuable introduction to how analysts can use transaction-level data to identify important customers, suppliers, and trading patterns.</p><h4>3. Applying My Skills to Nigeria’s Petroleum Trade</h4><p>To extend my learning beyond the provided examples, I also analysed Nigeria’s 2023 petroleum-product trade data from UN Comtrade, focusing on HS Code 2710.</p><p>This classification covers certain petroleum oils and related products, rather than crude petroleum alone.</p><p>I separated imports from exports, grouped transactions by reporting month, calculated total trading values, and identified the leading trading partners.</p><h4>My key findings</h4><p>For Nigeria’s recorded petroleum-product imports, January 2023 had the highest monthly value at approximately <strong>$2.33 billion</strong>, while June recorded the lowest at approximately <strong>$871.78 million</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/924/1*TrK2y20k-yyjSyLMzE_bEA.jpeg" /></figure><p>On the export side, October recorded the highest value at approximately <strong>$25.92 million</strong>, while March recorded the lowest at approximately <strong>$1.42 million</strong>.</p><p>I also identified the three leading import partners:</p><p>CountryApproximate import valueBelgium$5.29 billionNetherlands$2.83 billionIndia$2.35 billion</p><p>Together, these three countries accounted for approximately 52% of the recorded $20.11 billion in imports within the dataset.</p><p>These findings demonstrated how grouping and aggregation can help reveal the concentration of international trade among major partner countries.</p><p>However, I also learned an important lesson about interpretation: one year of observations is not sufficient to establish a recurring seasonal pattern.</p><p>A particular month may have unusually high trading activity, but that does not necessarily mean the same pattern repeats every year.</p><p>This reminded me that good data analysis requires both calculations and careful judgment.</p><h4>4. Turning Data into Visual Stories</h4><p>Another exciting part of Week 8 was learning how to visualise data using Matplotlib and Seaborn.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/863/1*vuXSSNWozg1XR0XiyxKmKg.jpeg" /></figure><p>I successfully created five different visualisations using the UK milk trade dataset.</p><h4>Histogram</h4><p>I created a histogram to understand the distribution of trade values.</p><p>A histogram groups numerical observations into ranges, helping analysts see where values are concentrated and how frequently different value ranges occur.</p><h4>KDE Plot</h4><p>The Kernel Density Estimate (KDE) plot provided a smoothed representation of the trade-value distribution.</p><p>This helped me understand another way of examining the overall shape of numerical data.</p><h4>Box Plot</h4><p>I created a box plot to examine the median, middle 50% of observations, spread, and potential outliers.</p><p>One thing I found useful was seeing how a simple visualisation can highlight unusually large or small observations that may deserve further investigation.</p><h4>Regression Plot</h4><p>The regression plot introduced me to visualizing relationships between numerical variables.</p><p>Using the reporting period and trade value, I learned how a fitted line can summarize an estimated relationship.</p><p>I also learned to interpret this cautiously because a fitted relationship does not automatically establish causation or describe changes in monthly totals.</p><h4>Heatmap</h4><p>Finally, I created a heatmap comparing monthly import and export totals.</p><p>The heatmap made it easier to represent numerical differences using colour intensity.</p><p>The broader lesson was that different charts serve different purposes. Choosing the appropriate visualisation depends on the question we are trying to answer.</p><p>I also practised adding descriptive titles, axis labels, and short interpretations so that someone viewing the chart could understand its purpose without reading the underlying code.</p><h4>5. Strengthening My SQL Skills</h4><p>One of the most rewarding parts of the assignment was returning to SQL after working extensively with Pandas.</p><p>I started seeing how operations I performed in Python could also be expressed directly in SQL.</p><p>For example, Pandas groupby() has a close relationship with SQL GROUP BY.</p><p>Both allow us to organise records into groups and calculate meaningful summaries.</p><p>The take-home assignment contained five SQL practice tasks, and I successfully completed all five using the provided automatic checker.</p><h4><strong>SQL Task 1 — Grouping by Two Columns</strong></h4><p>I used GROUP BY, COUNT(DISTINCT ...), and CASE to summarize the UK&#39;s 2015 milk trade by trade flow and product classification.</p><h4><strong>SQL Task 2 — Monthly Imports and Exports</strong></h4><p>I used conditional aggregation to calculate monthly imports, exports, and the trade balance for the 2014 data.</p><p>This exercise reinforced an important calculation rule: compute the trade balance using unrounded totals and round only the final result.</p><h4><strong>SQL Task 3 — Quarterly Reporting with GROUPING SETS</strong></h4><p>I learned how to generate quarterly summaries and annual subtotals in the same SQL query.</p><p>This was particularly useful because reporting at multiple levels is common in business analytics.</p><h4><strong>SQL Task 4 — Ranking Trading Partners</strong></h4><p>I used window functions, including ROW_NUMBER() SUM() OVER(), to rank the top three trading partners separately for imports and exports.</p><p>I also calculated each partner’s percentage contribution to its trade flow.</p><h4><strong>SQL Task 5 — Preparing Data for Visualisation</strong></h4><p>The final SQL task involved identifying the UK’s five largest export partners and grouping the remaining countries into an Other category.</p><p>I then prepared a tidy monthly summary containing export totals and percentage shares, suitable for a stacked bar chart.</p><p>Completing these exercises helped me understand that SQL is not limited to retrieving database records. It is also a powerful tool for performing analytical calculations and preparing data for reporting.</p><h4>6. Challenges I Encountered and How I Overcame Them</h4><p>This learning experience was not without challenges.</p><p>One difficulty was understanding how Pandas handles grouping keys, particularly when working with multiple columns. I encountered situations where selecting a group required tuple-based indexing, which initially felt confusing.</p><p>As I explored the outputs and reviewed the structure of the grouped data, the reason became clearer.</p><p>Another challenge involved commodity codes such as 0401 and 0402.</p><p>These codes contain leading zeros and must be handled as text. Treating them as integers can change their representation and cause filtering or grouping issues.</p><p>I also learned why data preparation matters before aggregation.</p><p>Some trade datasets include a World partner record representing totals across all countries. If these summary records are included alongside individual country records, the resulting calculations can double-count the trading values.</p><p>Separating summary records from country-level observations became an important part of my workflow.</p><p>In SQL, I needed to pay closer attention to grouping logic, column order, sorting, decimal precision, and rounding.</p><p>The automatic practice checker was particularly useful because it encouraged me to verify the actual results rather than assume a query was correct simply because it executed without an error.</p><p>Most importantly, I learned to approach errors systematically: understand the message, inspect the data or query, correct one issue at a time, and rerun the calculation.</p><h4>7. Lessons I Will Carry Forward</h4><p>This week changed how I think about data analysis.</p><p>I learned that the first step should always be understanding the dataset: what each row represents, what each column means, which values are missing, and whether summary records are mixed with detailed observations.</p><p>I also learned that analytical accuracy depends on using the correct aggregation method.</p><p>A monthly total, an average transaction value, and a maximum individual transaction may tell different stories, even when calculated from the same dataset.</p><p>Visualisation is equally important, but charts should support accurate interpretation rather than simply make the analysis look attractive.</p><p>Another takeaway is that SQL and Pandas complement each other. SQL can retrieve, filter, aggregate, and structure records directly in a database, while Pandas provides flexible tools for further exploration, transformation, and visualisation.</p><p>For anyone beginning a similar journey, my advice is to practice with real datasets, question every result, and focus on understanding the logic behind each line of code.</p><p>It is also worth documenting mistakes and the solutions that worked. Those experiences often become some of the most valuable learning moments.</p><h4>Conclusion: Building Confidence Through Practice</h4><p>Week 8 was an important milestone in my data analytics journey.</p><p>I progressed from grouping and aggregating records in Pandas to analyzing international trade, creating five different types of visualizations, and successfully completing five SQL practice exercises.</p><p>Working with UK milk trade and Nigerian petroleum-product trade data helped me connect technical concepts with real-world questions.</p><p>I am beginning to understand that becoming a good data analyst is not simply about memorising Python functions or SQL commands.</p><p>It is about developing the ability to turn data into evidence, communicate findings clearly, recognise limitations, and make informed conclusions.</p><p>I still have much more to learn, but every completed exercise is helping me build stronger analytical skills and greater confidence.</p><p>My job as a learner is to figure out how to make sense of messy data. And trust me, it isn’t always straightforward.</p><p><strong>My biggest takeaway from Week 8: Data becomes valuable when we understand how to organise it, analyse it, and explain what it means. It will take a while, but understanding is certain after the initial confusion.</strong></p><p>This is my documentation for my internship at DataraFlow, and you can get access the full notebook here on GitHub: <a href="https://github.com/Nneomadmapgirl">https://github.com/Nneomadmapgirl</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=00c15aac6cb8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SOURCING, CLEANING, AND ANALYSING DATA USING PYTHON PANDAS AND SQL]]></title>
            <link>https://python.plainenglish.io/sourcing-data-cleaning-and-analysing-using-python-pandas-and-sql-7a78454b4a24?source=rss-bbfcc9f38a59------2</link>
            <guid isPermaLink="false">https://medium.com/p/7a78454b4a24</guid>
            <category><![CDATA[python-libraries]]></category>
            <category><![CDATA[exploratory-data-analysis]]></category>
            <category><![CDATA[sql-server]]></category>
            <dc:creator><![CDATA[Nneoma Ezealuma (dmapgirl)]]></dc:creator>
            <pubDate>Sat, 24 Oct 2026 21:44:36 GMT</pubDate>
            <atom:updated>2026-11-11T00:43:13.069Z</atom:updated>
            <content:encoded><![CDATA[<h4>I sourced these datasets myself, and I feel super proud of my development so far👌✌</h4><p>My article for this week will be split into 4 parts, each detailing the process undertaken to achieve my results and conclusions.</p><p><strong>PART A</strong></p><p><strong>Sourcing the Weather Data, Cleaning and Analysing Using Python and Pandas</strong></p><p>This was my first time sourcing real-life data. Every dataset I have ever used was handed down to me until this week. I realised what “Dirty Datasets” actually means because I had to clean and merge this particular dataset until it was sparkling. I downloaded monthly data individually for London Heathrow, United Kingdom, pasted each month into my notepad, then converted it to a CSV file. This practical assignment was accomplished using Python and the Pandas library. The main objective was to load the file, clean and organise the weather data, and analyse it, particularly by extracting records for the summer months of June, July, and August which are the summer months.</p><p><strong>1. Loading and Understanding the Dataset</strong></p><p>I started by importing the CSV file with pd.read_csv() and examining its structure using .head(), .shape, and .dtypes. These helped me understand the number of rows and columns, view the first few records, and identify the data types.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1012/1*0IFz0wi272abPK7C30IrvQ.jpeg" /><figcaption>This is a screenshot from my notebook showing the dataset laoding process</figcaption></figure><p><strong>2. Cleaning and Correcting the Data</strong></p><p>Before I started cleaning, I had a sneak peek into this dataset, just to understand the data I am working with.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1005/1*xQ_gzS0vDpeNzJArIe5uCw.jpeg" /><figcaption>A screenshot of the loaded dataset</figcaption></figure><p>During the cleaning process, I discovered that the dataset contained duplicate-looking Date information, unnecessary unnamed columns, and two header rows. The second header row contained labels such as Max, Avg, and Min. To correct this, I reloaded the CSV using skiprows=2, header=None, and usecols=range(17), assigning appropriate column names to the weather measurements. This produced a properly structured dataset containing <strong>365 rows and 16 weather measurement columns</strong>, excluding the Date index.</p><p><strong>3. Converting Data Types and Setting the Date Index</strong></p><p>I learned how to convert date values into datetime objects using pd.to_datetime() and how to set the Date column as the DataFrame index using .set_index(‘Date’).</p><p>I also converted weather measurements from strings into numeric values using pd.to_numeric(). The errors=’coerce’ argument helped handle invalid entries. We also explored checking for missing values using .isnull().sum() and identifying duplicate dates with .duplicated().sum().</p><p><strong>4. Filtering Summer Weather Records</strong></p><p>The next task was to create a new DataFrame containing only the summer months in the Northern Hemisphere. I used the datetime index with .month.isin([6, 7, 8]) to identify June, July, and August, then applied Boolean filtering to select the relevant records.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2MoEdF8yTACUA3O4y9jixg.jpeg" /><figcaption>This notebook screenshot shows the filtering code block for the summer months</figcaption></figure><p>The earlier version of the dataset returned <strong>92 summer observations</strong>, representing the 30 days of June, 31 days of July, and 31 days of August.</p><p><strong>5. Performing Basic Weather Analysis</strong></p><p>I practiced using .mean(), .max(), and .min() to calculate averages and identify the highest and lowest temperatures. I also learned how to use .groupby() to compare temperatures across individual summer months and .idxmax() to identify the month with the highest average.</p><p>An important discovery was that the initial temperature calculations used daily maximum temperatures rather than daily average temperatures. Correcting the column structure allowed me to identify the proper measurements for future analysis.</p><p><strong>PART B AND C</strong></p><p><strong>Exploring World Bank Data Using SQL, Python, Pandas, and Statistical Analysis</strong></p><p><strong>Overview</strong></p><p>This part contains two projects, B and C, respectively. Throughout this learning session, I worked on two data science projects using World Bank datasets. The objective was to develop practical skills in retrieving, cleaning, transforming, merging, analysing, and visualising data while investigating relationships between socioeconomic indicators.</p><p><strong>Project B: GDP and Life Expectancy</strong></p><p>The first project examined the relationship between countries’ GDP and life expectancy in 2013. Using SQL, I retrieved two datasets from the cohort database, each containing 248 rows and three columns.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/521/1*RSMqR1udKBbngGhmlRmURg.jpeg" /><figcaption>Screenshot of my retrieving code block from my notebook</figcaption></figure><p>I learned how to inspect datasets using .head(), .tail(), and .shape, and identify missing values using .isna().sum(). The GDP dataset contained 23 missing values, while the life expectancy dataset contained 17. I explored removing missing values with .dropna() and excluding regional aggregates to focus on individual countries.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/559/1*NZYlKg3m0EiFIU13-amiEA.jpeg" /><figcaption>Notebook with code block for identifying missing values</figcaption></figure><p>I also practised data transformation by converting GDP from US dollars into millions of British pounds, rounding life expectancy values, and selecting relevant columns. Using pd.merge(), I learned how to combine datasets by matching country names. The project introduced Spearman correlation, scatterplots, and sorting data to investigate economic and health relationships.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/990/1*bCCmmrvDpmPyH6R3pR1mxA.jpeg" /><figcaption>Spearman Correlation from my notebook</figcaption></figure><p><strong>Project C: Electricity Access and Child Mortality</strong></p><p>For my independent project, I investigated whether greater access to electricity was associated with lower under-five mortality rates across countries in 2013.</p><p>I retrieved two World Bank datasets through an API, each containing 266 records. After removing missing values and regional aggregates, the electricity dataset contained 215 records, and the mortality dataset contained 196. An inner merge produced <strong>195 matched observations with no duplicate countries</strong>.</p><p>The Spearman correlation analysis produced:</p><ul><li><strong>Correlation coefficient:</strong> -0.8831</li><li><strong>P-value:</strong> approximately 2.39 × 10⁻⁶⁵</li><li><strong>Statistical significance:</strong> Significant at the 5% level</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/809/1*aXy5MT-n4ej5ZZLJ3qFTYg.jpeg" /><figcaption>Electricity and Child Mortality Scatter plot from my notebook</figcaption></figure><p>These findings revealed a strong negative relationship: countries with greater electricity access generally had lower under-five mortality rates.</p><p>Further exploration showed that Chad, Sierra Leone, and the Central African Republic had particularly low electricity access and high child mortality. However, Nigeria and Somalia demonstrated that relatively higher electricity access does not necessarily guarantee low mortality.</p><p><strong>Key Skills and Lessons Learned</strong></p><p>Through both projects, I developed a foundational understanding of:</p><ul><li><strong>Data retrieval:</strong> Using SQL queries and APIs to obtain datasets.</li><li><strong>Data cleaning:</strong> Identifying missing values, regional aggregates, and duplicate records.</li><li><strong>Data transformation:</strong> Renaming columns, converting units, and applying Python functions.</li><li><strong>Data merging:</strong> Combining related datasets using common country identifiers.</li><li><strong>Statistical analysis:</strong> Interpreting Spearman correlation coefficients and p-values.</li><li><strong>Exploratory data analysis:</strong> Creating scatterplots, sorting observations, and filtering records using conditions.</li></ul><p>I also learned that programming is less about memorising code and more about understanding its purpose, applying the appropriate functions, and verifying results.</p><p><strong>PART D SQL Practice</strong></p><p>During this SQL practice, I learned how to retrieve, clean, filter, combine, and analyse data directly from a database using SQL. Building on my previous experience with Python and pandas, I explored how SQL performs similar data analysis operations without first loading entire datasets into pandas.</p><p><strong>1. Setting Up the SQL Environment- It was quite daunting because I encountered a lot of errors</strong></p><p>I configured my Jupyter Notebook to connect to the course database. Initially, I encountered a ModuleNotFoundError because the practice.py file was not in the expected db folder. After correcting the file location, the setup ran successfully.</p><p>I learned that q() executes SQL queries for exploration, while p.sql() executes queries and checks assignment answers.</p><p><strong>2. Cleaning Messy Weather Data, Task 1 ✅</strong></p><p>Using the week07.london_2014 table, I cleaned wind direction values containing unwanted HTML text (&lt;br /&gt;).</p><p>I learned to use:</p><ul><li>SELECT and FROM to retrieve data.</li><li>REPLACE() to remove unwanted characters.</li><li>TRY_CAST() to convert text into integers.</li><li>WHERE to filter records.</li><li>COUNT(*) to count matching rows.</li></ul><p><strong>Result:</strong> Identified <strong>5 days</strong> with wind direction greater than 350 degrees. The assignment checker confirmed the answer.</p><p><strong>3. Working with Dates, Task 2</strong></p><p>I learned to convert dates stored as text into actual SQL dates using TRY_CONVERT(date, GMT). I also used BETWEEN to filter records within an inclusive date range and ORDER BY to sort them chronologically.</p><p><strong>Result:</strong> Retrieved five days of London weather observations, from December 8–12, 2014, including maximum and minimum temperatures.</p><p><strong>4. Combining World Bank Tables, Task 3 ✅</strong></p><p>I learned how INNER JOIN combines matching records from multiple tables using a common column. I joined the 2013 World Bank GDP and life expectancy datasets using country.</p><p>Additional concepts included AND, IS NOT NULL, ROUND(), TOP 10, and sorting with ASC and DESC.</p><p><strong>Result:</strong> Identified the 10 entries with the highest life expectancy among those with GDP below $10 billion. The Faroe Islands ranked highest at 81.3 years. The assignment checker confirmed the result.</p><p><strong>5. Identifying Missing Data, Task 4 ✅</strong></p><p>I learned the difference between INNER JOIN and LEFT JOIN. While INNER JOIN retains matching records, LEFT JOIN preserves all records from the left table.</p><p>Using IS NULL, OR, and COUNT(*), I identified records with missing GDP or life expectancy information.</p><p><strong>Result:</strong> Found <strong>29 unusable rows</strong> in the GDP dataset. The assignment checker confirmed the answer.</p><p><strong>6. Window Functions and Rolling Rainfall , Task 5</strong></p><p>I explored SQL window functions to calculate rolling seven-day rainfall totals during London’s summer of 2014.</p><p>Key concepts included SUM() OVER(), ROWS BETWEEN CURRENT ROW AND 6 FOLLOWING, COUNT() OVER(), and Common Table Expressions (WITH).</p><p>I learned that a seven-day rolling window moves forward <strong>one day at a time</strong>. For example, the first calculation covers June 1–7, the second June 2–8, and the third June 3–9. Each result is attached to the starting date.</p><p>I also learned how to exclude incomplete windows and use TOP 1 with ORDER BY to identify the driest period. The final assignment check is still pending.</p><p>SUMMARY AND CONCLUSION</p><p>These projects helped me understand the complete data science workflow, from obtaining raw data to interpreting statistical findings. Project B introduced economic and health comparisons, while Project C demonstrated a strong relationship between infrastructure access and child survival.</p><p>The most important lesson was that <strong>correlation does not imply causation</strong>. Although electricity access and child mortality were strongly associated, healthcare, nutrition, sanitation, education, and economic conditions may also influence the relationship.</p><p>The SQL practice strengthened my understanding of SQL as a tool for real-world data analysis. I can now perform fundamental data cleaning, filtering, joining, aggregation, missing-value analysis, and rolling calculations. Most importantly, I learned the relationship between SQL and pandas: <strong>SQL can prepare and analyse data inside a database, while pandas provides additional flexibility for analysis in Python.</strong></p><p><strong>One important thing I realized in all this is that where and how you store your files matters because, if Python cannot access them, then it throws an error. I encountered this because I wrongly saved a single file, and I rectified it with patience. So please, be patient with yourself if you are learning like me; gradually, it will start making sense, and don&#39;t try to memorise code; understand the functions and their applications.</strong></p><p>This is still my Dataraflow internship documentation, and for my complete notebook, you can access it here: <a href="https://github.com/Nneomadmapgirl">https://github.com/Nneomadmapgirl</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7a78454b4a24" width="1" height="1" alt=""><hr><p><a href="https://python.plainenglish.io/sourcing-data-cleaning-and-analysing-using-python-pandas-and-sql-7a78454b4a24">SOURCING, CLEANING, AND ANALYSING DATA USING PYTHON PANDAS AND SQL</a> was originally published in <a href="https://python.plainenglish.io/">Python in Plain English</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From Hidden Spaces to Missing Countries: Wrestling with NumPy, Pandas, SQL, and Global Weather Data]]></title>
            <link>https://python.plainenglish.io/from-hidden-spaces-to-missing-countries-wrestling-with-numpy-pandas-sql-and-global-weather-data-2f51856acbea?source=rss-bbfcc9f38a59------2</link>
            <guid isPermaLink="false">https://medium.com/p/2f51856acbea</guid>
            <category><![CDATA[sql-server]]></category>
            <category><![CDATA[python-libraries]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Nneoma Ezealuma (dmapgirl)]]></dc:creator>
            <pubDate>Mon, 19 Oct 2026 10:31:45 GMT</pubDate>
            <atom:updated>2026-10-19T10:31:45.618Z</atom:updated>
            <content:encoded><![CDATA[<p>It felt like a proper volcanic eruption in my brain, with lava flowing everywhere possible. One moment you’re writing element-wise comparisons in NumPy, the next you’re debugging joins, chasing running totals, and wondering why twenty-eight countries vanished from your report because of a tiny naming mismatch.</p><p>This week I bridged the gap between local data manipulation in Python (NumPy and Pandas) and querying structured data directly in a live Microsoft SQL Server database. I profiled tables, analysed 2014 climate data from six major airports, and worked through an intermediate SQL assessment. When exploring an unfamiliar database, a lot of steps are taken to avoid running into errors when analysing the data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1005/1*dIavJ7xoC7QpZp5CFTh2Tw.jpeg" /><figcaption>This screenshot from my notebook shows a block of code I used to know what tables exist, what columns do they have, and what the data actually looks like</figcaption></figure><h4>Database exploration comes first, no matter what.</h4><ol><li>System catalogues are your map. INFORMATION_SCHEMA.COLUMNS and INFORMATION_SCHEMA.TABLES let you inspect a database’s structure without opening individual tables. You can search for columns like “Income” or “Age”, or list every table across schemas.</li><li>Profile before you model. Before building anything, I check a table’s health with COUNT(*), COUNT(DISTINCT …), MIN, and MAX. Together they show the dataset’s scale, its category diversity, and any unrealistic values.</li><li>Peek smartly. Senior data scientists use SELECT TOP (or .head() in Pandas) to look at a small snapshot. On warehouses with billions of rows, pulling a whole table just to look at it can crash your machine and rack up serious cloud costs.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dTcaQKFMLQyYK0fasclWqg.jpeg" /><figcaption>This screenshot shows every column of the weather table I am peeking into, with its type, in table order</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/561/1*O4_DAaHyjBV81xq1xR_nMw.jpeg" /><figcaption>I used this block of code to SELECT TOP 5 rows of the table. It only return the very first 5 rows of data</figcaption></figure><h4>SQL techniques</h4><ol><li>Conditional aggregation with CASE WHEN to build custom categories and calculate metrics like churn and approval rates.</li><li>Group filtering with HAVING.</li><li>Window functions like ROW_NUMBER() OVER (…) and SUM(…) OVER (…) to rank temperatures and compute year-to-date running totals.</li><li>Wildcards with LIKE ‘%text%’, a flexible net that catches terms despite typos or hidden trailing spaces, where = fails.<br>Safe, read-only querying, with _BLOCKED filters to protect the database while exploring schemas.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/682/1*lUeIdRapRMQYJmnhf89gxA.jpeg" /><figcaption>I used this block of code to check for blank spaces in the data</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/579/1*gzZ6D0bCehDoOLbWraO8BQ.jpeg" /><figcaption>Instead of clicking through every table manually, I used this block of code to query and find a column anywhere in the database</figcaption></figure><p><strong>NumPy and Pandas: the local toolkit</strong><br>1. NumPy thinks in arrays. Element-wise comparisons and evaluations apply to a whole array at once, with no loops. I also tested how NumPy handles complex numbers.</p><p>2. Pandas Series and DataFrames are the bridge. I practised converting plain Python lists into Series, then moved on to DataFrames, which is where SQL results land.</p><p>3. Three methods give a first look at any DataFrame. .shape tells you the size, .info() shows column types and non-null counts, and .describe() gives summary statistics. It’s the Pandas version of profiling a table.</p><p>4. .head() is the Pandas SELECT TOP. Look at a small snapshot before doing anything else.</p><p>Python code below</p><p>import numpy as np<br>import pandas as pd</p><p>temps = np.array([-4, 3, 12, 25, 30])<br>print(temps &gt; 10) # element-wise: [False False True True True]</p><p>s = pd.Series([-4, 3, 12, 25, 30]) # list -&gt; Series<br>print(s.describe())</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/509/1*yCQXwNagiKFaLjphkgx8TA.jpeg" /></figure><p><strong>What I Solved</strong><br>1. Warm-ups in NumPy and Pandas</p><p>Before touching the database, I worked through element-wise evaluations, complex number tests, and list-to-Series conversions. These small tasks built the confidence for the bigger ones, and the habits carried straight over: the .shape / .info() / .describe() routine in Pandas mirrors the COUNT / MIN / MAX routine in SQL.</p><p>2. The 2014 airport weather analysis</p><p>I queried tables for Beijing, Brasilia, Cape Town, Delhi, London, and Moscow, pulled them into Pandas, and built a climate profile:</p><ul><li>Data integrity: Every city had exactly 365 daily rows for 2014, so no days were dropped.</li><li>Coldest: Moscow hit the year’s lowest temperature at -26°C, with a 59-degree annual swing.</li><li>Hottest: Beijing recorded the highest absolute temperature at 42°C.<br>Most stable: London had winter lows of only -4°C and a summer high of 30°C.<br><strong>Advanced SQL tasks</strong><br>1. Extracted schema properties and counted missing events across multiple cities.<br>2. Computed conditional churn percentages by contract type in a customer database.<br>3. Used anti-joins to isolate mismatched records between two global databases.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/813/1*fjRtzBWb6t7R5ZVoEi2sPg.jpeg" /><figcaption>A screenshot from my advanced SQL notebook</figcaption></figure><h4><strong>Challenges and How I Overcame Them</strong><br>Challenge 1: The sneaky space trap</h4><p>The problem: Many exported columns had accidental spaces in their names, like [ Events], [ CloudCover], and [ Max TemperatureC]. My queries kept failing with “column not found” errors.</p><p>The fix: I ran a meta-query against INFORMATION_SCHEMA.COLUMNS with LIKE ‘% %’ to find every column with hidden spaces, then wrapped them in brackets:</p><p>sql<br>SELECT COLUMN_NAME<br>FROM INFORMATION_SCHEMA.COLUMNS<br>WHERE COLUMN_NAME LIKE ‘% %’;</p><p>SELECT [ Events], [ CloudCover]<br>FROM weather_table;</p><p>The same trap exists in Pandas, where df[‘ Events’] and df[‘Events’] are different columns. Once the data was in a DataFrame, df.columns.str.strip() cleaned it up.</p><h4><strong>Challenge 2: The 28 missing countries</strong></h4><p>The problem: An inner join between the WHO table (week06.who_pop_tb_all) and the World Bank table (week08.wb_pop_2013) silently dropped 28 countries.</p><p>The fix: I wrote an anti-join (a LEFT JOIN filtered for NULLs on the right side) to list exactly which names didn’t match:</p><p>sql<br>SELECT w.country<br>FROM week06.who_pop_tb_all AS w<br>LEFT JOIN week08.wb_pop_2013 AS b<br> ON w.country = b.country<br>WHERE b.country IS NULL;</p><p>The culprit was naming. Different organizations name the same place differently, like “United States” versus “United States of America”. For real projects, the lasting fix is a standard mapping table. (Pandas has the same check: pd.merge(…, how=’left’, indicator=True) flags which rows found no match.)</p><h4><strong>Challenge 3: The integer-division trap</strong></h4><p>The problem: AVG([Mean TemperatureC]) and other divisions returned truncated whole numbers.</p><p>The fix: The underlying columns were integers, so SQL used integer division. Multiplying by 1.0 (or 100.0 for percentage rates) before dividing forces decimal arithmetic:</p><p>sql<br>SELECT 100.0 * SUM(CASE WHEN churned = 1 THEN 1 ELSE 0 END) / COUNT(*) AS churn_rate<br>FROM customers;</p><p>Python and NumPy behave differently here: / always gives a float, and // is explicit integer division. Knowing each tool’s defaults is half the battle.</p><p>I will still update this process progressively, but for now, let me get back to my notebook. Thanks for following my data science journey.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2f51856acbea" width="1" height="1" alt=""><hr><p><a href="https://python.plainenglish.io/from-hidden-spaces-to-missing-countries-wrestling-with-numpy-pandas-sql-and-global-weather-data-2f51856acbea">From Hidden Spaces to Missing Countries: Wrestling with NumPy, Pandas, SQL, and Global Weather Data</a> was originally published in <a href="https://python.plainenglish.io/">Python in Plain English</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From Prediction to Protection: Using Machine Learning and Maps to Fight Flash Floods in Abuja]]></title>
            <link>https://dmapgirl.medium.com/from-prediction-to-protection-using-machine-learning-and-maps-to-fight-flash-floods-in-abuja-0edc99a51385?source=rss-bbfcc9f38a59------2</link>
            <guid isPermaLink="false">https://medium.com/p/0edc99a51385</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[mapping]]></category>
            <category><![CDATA[gis]]></category>
            <category><![CDATA[infinite-library]]></category>
            <category><![CDATA[environmental-issues]]></category>
            <dc:creator><![CDATA[Nneoma Ezealuma (dmapgirl)]]></dc:creator>
            <pubDate>Sat, 10 Oct 2026 00:25:34 GMT</pubDate>
            <atom:updated>2026-10-10T20:12:24.052Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/720/1*GaUn61H71tVuQ8be3WX-Bg.jpeg" /><figcaption>Nneoma Dmapgirl during a field work at one of the gully sites caused by heavy flooding, endangering the buildings visible in the photo</figcaption></figure><p>Every single day, hundreds of people pack up their bags and move to Abuja. The city is booming, estates are springing up overnight, and concrete roads are stretching across areas that used to be wide-open green fields. But this rapid growth hides a quiet, dangerous problem: when the heavy rains hit the Federal Capital Territory (FCT), the water has nowhere to go.<br>Instead of soaking into the soil, it pools on asphalt and concrete, turning streets into raging rivers and catching entire neighbourhoods completely off guard.<br>When we see devastating floods on the news in Nigeria, the focus usually lands on coastal cities like Lagos or major river confluences like Lokoja. But Abuja’s unique urban flash flooding is turning into a massive crisis.<br>As part of my data science journey during the DataraFlow internship, I’m spending the next six months building a project to change how we tackle this problem. The research is titled: <strong>“FROM PREDICTION TO PROTECTION: A Geospatial Machine Learning Approach to Flood Risk Prediction and Community-Level Awareness in Abuja.”</strong><br>Here is exactly how I plan to use satellite data, machine learning, and basic cartography to bridge the gap between complex data and real-world human safety.</p><h3><strong>The Real Problem: We Are Too Reactive</strong></h3><p>Right now, our approach to flood management in Abuja is almost entirely reactive. We tend to wait until a neighbourhood is submerged, and then we send in emergency management teams to rescue stranded residents and clean up the aftermath. The weather alerts available to the public are usually too broad or arrive too late. A generic radio or TV announcement saying “Expect heavy rainfall in the FCT” doesn’t help a family living in an informal settlement or a low-lying valley figure out if their specific street is going to be underwater in three hours. Because vulnerable communities don’t have access to clear, localised hazard data, they can’t take proactive, sustainable measures like cleaning out specific clogged drainage systems, reinforcing their property boundaries, or organising neighbourhood evacuation paths before the wet season peaks.<br>If we keep keeping this data locked inside complicated spatial software, nothing changes. We need to take advanced environmental data, pass it through smart predictive models, and then translate those outputs into simple tools that everyday people can actually use.</p><h3>The Plan: Building a Data-Driven Workflow</h3><p>Over the next six months, I am merging together a step-by-step digital workflow that turns raw environmental data into a life-saving community early warning framework.<br><strong>My project is broken down into three major steps:</strong></p><p><strong>1. Feeding the Engine (The Environmental Factors)</strong></p><p>To predict where water will pool, a computer needs to understand the terrain. I will be collecting, cleaning, and overlaying distinct environmental layers across a digital map of Abuja:<br><strong>• Weather Data: </strong>Tracking historical rainfall distribution to find where the heaviest downpours consistently land.<br>•<strong> The Shape of the Land:</strong> Using 30-meter elevation models to calculate terrain slopes, ridges, and low-lying basins.<br><strong>• The Waterways</strong>: Mapping how close neighbourhoods are to rivers and calculating the density of local drainage gutters.<br><strong>• The Surface and Earth:</strong> Examining soil properties, geological rock types, and satellite greenness indexes (NDVI) to see how well the ground absorbs rushing water. <strong>Of course, this is essential; I am a geologist.</strong><br><strong>• Human Footprints:</strong> Mapping out exactly where urban concrete expansion has paved over natural floodplains.</p><p><strong>2. Training the Machine Learning Models</strong></p><p>Once I have a complete inventory of points where Abuja has flooded in the past, I will pair them with these environmental factors. Using the skills I am learning in my data science track, I will train classification algorithms like Random Forest and Support Vector Machines (SVM).<br>By splitting the data (70% to train the model and 30% to validate its accuracy), I can test which algorithm is the absolute best at predicting future flood locations based on terrain and weather patterns.</p><p><strong>3. Creating the Map (Moving Data to Design)</strong></p><p>Once the machine learning model gives me its raw mathematical probabilities, I will step in with my core skill: map creation.<br>Using GIS software, I will translate those numbers into a beautifully styled, high-resolution Flood Risk Map of Abuja. The map will categorise the FCT into five clear, colour-coded danger zones: Very Low, Low, Medium, High, and Very High Risk. Anyone from a city planner to a local chief will be able to look at it and understand the danger level of their area at a single glance.</p><p><strong>The Ultimate Goal: Turning Math into Local Action</strong></p><p>An academic map sitting on a data scientist’s laptop doesn’t save lives. <strong>The real heart of this project is the “Protection” phase.</strong><br>In the final months of the internship, I want to establish a framework that bridges the gap between science and community awareness. I want to use our predictive maps to isolate the highest-risk communities and build a template for local-language text alerts.<br>Imagine taking a high-probability flood prediction and automatically generating simple, clear short-text updates written in Hausa, Gbagyi, and Pidgin English that can be broadcast directly to community heads and residents.<br>This moves the conversation away from standard disaster rescue and pushes it toward community resilience and sustainable climate adaptation.</p><p><strong>My Core Sources</strong></p><p>To make sure this workflow is scientifically sound, I am grounding my research in real, peer-reviewed studies conducted right here in Nigeria. I am drawing inspiration from authors who have mapped flood susceptibility in the Ibadan Metropolis and the lower Niger River Basin, alongside specific local studies focusing on the Abuja Municipal Area Council (AMAC). These papers provide the mathematical justification I need to prove that my environmental layers are weighted correctly.<br>This project is a massive undertaking, but it represents exactly why I got into data science: using technology not just to analyse numbers, but to solve real, urgent problems happening in my own backyard.</p><p>I will see you soon with every tiny update I have to share; until then, you be good. Bye! From dmapgirl, see ya.</p><p>Write for Infinite Library: Submission Guidelines</p><p><a href="https://medium.com/infinite-library/infinite-library-a-publication-34773d0283a7">Infinite Library — A Publication</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0edc99a51385" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Learning Python and SQL Feels Confusing, But The Journey Continues at DataraFlow]]></title>
            <link>https://dmapgirl.medium.com/here-comes-learning-python-and-sql-when-everything-feels-confusing-3e860ef02426?source=rss-bbfcc9f38a59------2</link>
            <guid isPermaLink="false">https://medium.com/p/3e860ef02426</guid>
            <category><![CDATA[python-for-beginners]]></category>
            <category><![CDATA[data-science-courses]]></category>
            <category><![CDATA[sql-for-beginners]]></category>
            <category><![CDATA[python-for-data-science]]></category>
            <dc:creator><![CDATA[Nneoma Ezealuma (dmapgirl)]]></dc:creator>
            <pubDate>Fri, 25 Sep 2026 00:47:21 GMT</pubDate>
            <atom:updated>2026-09-25T01:02:18.519Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Learning Python and SQL Feels Confusing, But The Journey Continues at DataraFlow</strong></p><p>A beginner’s tour of modules, JSON, error handling, virtual environments, and SQL, in plain English, with small practice tasks you can try yourself.</p><p>I’ll be honest, when I started learning Python, almost everything confused me. Modules, JSON, virtual environments, error handling. Every lesson introduced a new command, and I was memorising them without knowing why they mattered. Do I know why they mattered now? Well, come along and find out.</p><p>I made a decision. Instead of waiting to understand everything first, I would practise my way into understanding. I type the code, change it, break it, and fix it. Hold on, some are still unfixed, but I will figure it out.</p><p>This DataraFlow lesson is where Python and SQL finally met, and the pieces snapped together.</p><p>If you’re a total beginner and feel lost too, this post is for you. Below is what I learned in simple terms, with everyday examples and a small “Try it yourself” task after each topic. Don’t just read it. Open your editor and type along. It can be challenging, but you can always start over from “Hello World”.</p><h3><strong>MODULES</strong></h3><p>Modules = Write Code Once, Use It Everywhere</p><p>A module is a Python file full of useful code that you can borrow in other programs.</p><p>Think of a recipe book. You don’t rewrite your pancake recipe every morning; you open the book and use it.</p><p>Save this in a file called mymodule.py:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*WDxpMB9D-JBP9DYTrt8Xxg.jpeg" /><figcaption>Here’s a code for importing the math module with it’s output</figcaption></figure><p>You can give a module a shorter nickname (an alias), or import only the piece you need. Good code isn’t only code that works. It’s code that is organised so you can reuse it.</p><h3><strong>DATE AND TIME</strong></h3><p>Almost every real system deals with dates: bank transactions, appointments, delivery times, website logs. Python’s datetime module handles them.</p><p>import datetime</p><p>x = datetime.datetime.now() returns current date and time<br>print(x)<br>print(x.year) returns just the year<br>print(x.strftime(“%A”)) returns the day of the week, e.g. “Thursday”</p><p>You can also create a specific date:<br>x = datetime.datetime(2020, 5, 17)</p><p>strftime() turns a date into readable text, like choosing whether a form shows “17 May 2020” or just “Sunday”.</p><p>Data isn’t only numbers and words. Time is data too.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/977/1*7CrPn2RyXPfF1AfgzfhFWw.jpeg" /><figcaption>Datetime code with its output</figcaption></figure><h3><strong>MATHS</strong></h3><p>Python has built-in helpers:</p><p>min() gives the smallest value<br>max() gives the largest value<br>abs() removes the minus sign<br>pow() raises a number to a power</p><p>The math module adds more:<br>import math</p><p>print(math.sqrt(64)) gives 8.0<br>print(math.ceil(1.4)) gives 2 (rounds UP)<br>print(math.floor(1.4)) gives 1 (rounds DOWN)<br>print(math.pi) gives 3.14159…</p><p>These look basic, but they are the building blocks of statistics, analytics, and machine learning.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1021/1*Fvs-uzu0xpjvmnSsF3Nmcg.jpeg" /><figcaption>Math operation and output</figcaption></figure><h3><strong>JSON</strong></h3><p>JSON is a Common Language for Sharing Data. It is a text format that apps and websites use to send data to each other. It looks a lot like a Python dictionary, like a form with labels and answers (“name: John, age: 30”).</p><p>import json</p><p>x = ‘{“name”:”John”, “age”:30, “city”:”New York”}’ These are JSON text</p><p>y = json.loads(x) returns text -&gt; Python dictionary<br>print(y[“age”]) returns 30</p><p>back_to_json = json.dumps(y) returns dictionary -&gt; text</p><p>A quick memory trick: loads loads a string into Python, and dumps dumps Python out as a string.</p><p>JSON is the bridge between Python and the outside world: APIs, cloud apps, and settings files all use it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/653/1*PWLYHUUDYQwxNmg1tAzKeQ.jpeg" /><figcaption>JSON string, conversion and ouput</figcaption></figure><h3><strong>ERRORS</strong></h3><p>Expecting Errors (Exception Handling)</p><p>A program that works when everything goes right is good. A program that knows what to do when something goes wrong is much better. Think of a phone charger with a safety cut-off.</p><p>try runs the code you want to test<br>except runs if there is an error<br>else runs if there is no error<br>finally runs no matter what</p><p>try:<br> print(x)<br>except NameError:<br> print(“Variable x is not defined”)<br>except:<br> print(“Something else went wrong”)<br>else:<br> print(“Everything worked!”)<br>finally:<br> print(“This always runs.”)</p><p>Don’t just ask, “Will my code work?” Ask, “What should my code do when something fails?”</p><h3><strong>Virtual Environments</strong></h3><p>A Separate Toolbox for Each Project</p><p>A virtual environment gives each project its own private set of Python packages. Imagine two cooks sharing one kitchen. One needs sugar, the other salt, and they keep mixing up the jars. Give each cook their own kitchen and the problem disappears. In the same way, Project A can use version 1 of a package while Project B uses version 2, without breaking each other.</p><p>Create one:</p><p>python -m venv myfirstproject</p><p>Activate it:</p><p>Windows: myfirstproject\Scripts\activate<br>Mac/Linux: source myfirstproject/bin/activate</p><p>You’ll see (myfirstproject) in your terminal. Now install a package just for this project:</p><p>pip install cowsay</p><p>pip is like an app store for Python packages.</p><p>Try it yourself: Create two environments with different names, install cowsay in only one, and see what happens when you try to import it in the other. That “it’s not installed here” error is the whole lesson.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/776/1*nWE0I26ggAuj5gKFDnhguA.jpeg" /><figcaption>The terminal showing myfirstproject and a successful pip install</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/547/1*JOYnLo0PtdpEvh1WCOW5Cg.jpeg" /><figcaption>The terminal after I successfully deactivated the virtual environment</figcaption></figure><h3><strong>Reading and Writing Files</strong></h3><p>Python opens files with open(). The mode tells Python what you want to do:</p><p>r reads the file<br>a appends (adds to the end)<br>w writes (replaces the contents)<br>x creates a new file (and errors if it already exists)</p><p>The safest way to open a file is with:</p><p>python<br>with open(“demofile.txt”) as f:<br> print(f.read())</p><p>with closes the file for you, even if something goes wrong. It’s like a door that shuts itself behind you.</p><p>Then Came SQL</p><p>Here’s the biggest shift. Until now, my datasets arrived as CSV files. In this stage, the data lives in a cloud database, and I have to ask for what I need using SQL.</p><p>That matters because reading a CSV is easy, but knowing how to ask a database for exactly the right data is a core data science skill.</p><p>A relational database is a collection of tables made of rows and columns, like a giant Excel workbook living on a server. SQL is how you ask it questions. I am yet to get a full grasp of SQL, so I will be dedicating a whole new write-up where I will explain vividly what SQL is all about from my understanding.</p><h3><strong>KEY NOTES</strong></h3><p>This stage wasn’t about memorising commands. It was about trying to fix the pieces myself.</p><p>Modules taught me reuse.<br>Datetime taught me that time is data.<br>JSON taught me how data is shared.<br>Exception handling taught me to expect failure.<br>Virtual environments taught me to keep projects separate.<br>File handling taught me how to save and read data.<br>SQL taught me how to ask a database for exactly what I need.</p><p>I still feel a little confused. I’m not going to pretend I understood all of this on the first read. I didn’t. What helped me was practice, not more reading, and I am going to practice even more, starting from the very beginning of Python.</p><p>Thirty focused minutes beat one long weekend session, and it’s okay to keep a “confusion list.” Some items solve themselves a week later, and the rest become great questions to ask.</p><p>Confusion isn’t a sign you’re failing. It’s what learning feels like before that sudden surge of high understanding</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3e860ef02426" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[PYTHON CRASH COURSE]]></title>
            <link>https://dmapgirl.medium.com/week-2-at-dataraflow-08acce443537?source=rss-bbfcc9f38a59------2</link>
            <guid isPermaLink="false">https://medium.com/p/08acce443537</guid>
            <dc:creator><![CDATA[Nneoma Ezealuma (dmapgirl)]]></dc:creator>
            <pubDate>Sat, 19 Sep 2026 04:54:48 GMT</pubDate>
            <atom:updated>2026-11-12T21:45:10.347Z</atom:updated>
            <content:encoded><![CDATA[<h4>Week 2 at DataraFlow</h4><p>It felt like Python looked at me and said, <em>“Okay, you know the basics… now let’s see what you can really do.”</em> 😄🐍</p><p>This week moved beyond simple variables and syntax into Object-Oriented Programming, inheritance, iterators, polymorphism, scope, and more advanced class behaviour. Suddenly, I wasn’t just writing lines of code — I was thinking about how different parts of a program could work together.</p><p>One of my favorite exercises was building an <strong>EvenNumbers iterator</strong> using __iter__() and __next__(). Seeing Python automatically generate 2, 4, 6, 8, 10 and stop with StopIteration made iterators finally click for me. I also built a <strong>Fibonacci iterator</strong>, which showed me how an object can actually keep track of its own state as it runs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/974/1*ih6w8ieU6q2AUUmSumbzkg.jpeg" /><figcaption>This is a screenshot of my notebook on fibonacci iterator. It’s basically about building a custom iterator and seeing StopIteration in action.</figcaption></figure><p>Then came <strong>inheritance and polymorphism</strong>. I worked with parent and child classes like Animal and Dog, and even multilevel inheritance with LivingBeing → Animal → Bird. The big lesson? Write useful behaviour once, inherit it where needed, and override it when something needs to behave differently.</p><p>My Cat and Dog classes also made polymorphism surprisingly simple:</p><p>animal_sound(Cat()) # Meow</p><p>animal_sound(Dog()) # Woof</p><p>Same function. Different objects. Different behaviour. Python basically said, <em>“I don’t care who you are, just bring the right method.”</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/616/1*ZQNM-bxsV67ESurZe9cwfQ.jpeg" /><figcaption>Screenshot of Polymorphism in action: same method, different behaviour.</figcaption></figure><p>Scope was another important stop. Using <strong>global</strong> and <strong>nonlocal</strong> helped me understand where variables live and how functions can modify variables outside their immediate environment. I also explored <strong>multiple inheritance and Method Resolution Order (MRO)</strong> with Flyer, Swimmer, and Duck, which showed me how Python decides which inherited method gets called first.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/829/1*nWtTAtrHPciR1jcHnsXTYw.jpeg" /><figcaption>Screenshot from my notebook exploring multiple inheritance and Python’s Method Resolution Order.</figcaption></figure><p>The real test was putting several concepts together in a <strong>Library Management System</strong>. I created different types of books, student and teacher members with different borrowing limits, iterable collections, private attributes, class variables, inheritance, and method overriding. That exercise made OOP feel much less theoretical; it started looking like something I could actually use to model real systems.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/682/1*ELlIE5vFXZdnPqBLle86AA.jpeg" /><figcaption>This is a part of the library management sytem concept</figcaption></figure><p>By the end of the week, Python had moved from:</p><p><strong>“Make this code work.”</strong></p><p>to:</p><p><strong>“How should this code be designed?”</strong></p><p>And that feels like an important shift.</p><p>Week 2 definitely stretched my brain out, but concepts like inheritance, iteration, polymorphism, and scope are starting to feel less like intimidating Python vocabulary and more like tools in my programming toolbox.</p><h3><strong>Data Structures and Algorithms</strong></h3><p>The week ended with a comprehensive YouTube tutorial on <a href="https://www.youtube.com/watch?v=kQDxmjfkIKY">Data Structures and Algorithms by CodDevX</a>. The two-hour video was demanding but worthwhile, and it strengthened my grasp of:</p><ul><li><strong>Built-in data structures:</strong> lists, tuples, sets, and dictionaries, including their pros and cons and when each one fits best.</li><li><strong>Core algorithm concepts:</strong> searching, sorting, and recursion.</li><li><strong>Useful Python techniques:</strong> slicing, comprehensions, and set operations.</li><li><strong>Practical implementations:</strong> stacks, queues, and heaps.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/353/1*vLbkQZPGU0u6xIXbfjKdgQ.jpeg" /><figcaption>A screenshot of a recursive factorial function that expects a non-negative integer</figcaption></figure><h3><strong>IN CONCLUSION</strong></h3><p>Week 2 was a whirlwind that almost engulfed me. Objects that run themselves, inheritance passing down the good stuff, polymorphism working its magic, and variable scope keeping everything in its lane. Then came data structures, and suddenly my code had a whole new rhythm.</p><p>Now, let’s be honest: one concept and I did <strong>NOT</strong> hit it off. It was difficult, it was new, and it looked at me like I was the one who didn’t belong. We didn’t part as friends. We parted like two people who agreed to “just see other tutorials.” 😅 But the syllabus waits for no one, so I waved goodbye, promised to revisit it later, and kept pedalling.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=08acce443537" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[MY FIRST STEPS TO INFINITY]]></title>
            <link>https://dmapgirl.medium.com/my-first-steps-to-infinity-bd3f4b4b91e3?source=rss-bbfcc9f38a59------2</link>
            <guid isPermaLink="false">https://medium.com/p/bd3f4b4b91e3</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[machine-learning-ai]]></category>
            <category><![CDATA[python3]]></category>
            <category><![CDATA[gis]]></category>
            <dc:creator><![CDATA[Nneoma Ezealuma (dmapgirl)]]></dc:creator>
            <pubDate>Thu, 10 Sep 2026 12:39:07 GMT</pubDate>
            <atom:updated>2026-09-10T12:39:07.918Z</atom:updated>
            <content:encoded><![CDATA[<p>(DATARAFLOW INTERNSHIP)WEEK 1</p><p>One Tiny Step at a Time</p><p>What small beginner exercises taught me about logic, patience, and actually understanding code……….Do I understand? Well, we will see in 3 months</p><p>My Python story started with one tiny line: print(“Hello, World!”). Small and Simple. But it was the start of learning to talk to a computer. I know it was for you too😂😂😂😂. I quickly learned something that changed how I studied: Python sticks better when you attach it to a small, real task instead of trying to swallow the whole language at once. You also have to be willing to accept the confusion that comes with not fully understanding what you are doing. Trust me, it’s confusing at first, but we will navigate it.</p><p>I stopped chasing everything and started chasing one idea at a time. That one habit took me from plain syntax to real tiny programs, ones that could decide, repeat, remember, break, and fix themselves.</p><h3>Where It All Began: The Basics🐌</h3><p>Every strong thing needs a floor to stand on. Mine was:</p><p>• Storing stuff in variables</p><p>• Grabbing input from a person with input()</p><p>• Showing results with print()</p><p>• Getting to know Python’s basic data types: numbers, floats, text, true/false, lists, tuples, dicts, and sets. With if, elif, and else, I built small tools that could……………</p><p>Nothing fancy. Just the raw pieces every program is made of. The code felt less like a script and more like a tiny brain making calls. Loops: Teaching Python to repeat itself so I don’t have to. Loops taught me programming automation, although I am yet to fully grasp but I get the idea. Using for loops, I built Multiplication tables, Factorial calculators, Fibonacci sequences, and the classic FizzBuzz challenge. Then I used a while loop for a number-guessing game, the kind where Python picks a secret number and you guess until you nail it. That one exercise quietly used loops, conditions, input, and the random module all at once. Lists, Tuples, and Dictionaries are where data gets real. Once I had single values under control, I moved to groups of them. With lists, I learned to store a bunch of values, find the biggest number, find the second biggest number, and pull out only the even ones.</p><p>I also learned the difference between a list and a tuple: a list can change, a tuple can’t. Then came dictionaries, pairing a key with its value. I used one to build a shopping cart, storing item names next to their prices, then adding it all up. I wrote a function that took two numbers and gave back their sum, difference, and product. I wrote another that filtered even numbers out of a list. I realised that functions make code easier to read, reuse, and fix.</p><h3>The Sneaky Concepts That Trip Beginners Up😔😔</h3><p>A few ideas don’t shout for attention, but they matter a lot as programs grow, such as local vs. global variables, where a variable lives and who can see it. == vs. is: checks if values match; the other checks if they’re the same object. If <strong>name</strong> == “<strong>main</strong>”: is the quiet gatekeeper that controls when code actually runs.</p><p>None of these looks exciting on paper. Sometimes code fails, so I practised catching them with try and except, specifically, stopping a program from crashing when someone divides by zero, and showing a kind message instead. Small fix, big lesson. It handles the mess. Making Data Stick Around.</p><p><strong>What It All Adds Up To</strong></p><p>Looking back, these small exercises quietly covered Python’s core: variables, input and output, data types, conditions, loops, functions, lists, tuples, dictionaries, strings, modules, errors, files- basically the bulk of the foundation. Each tiny program taught one new thing while quietly reinforcing everything before it, such as: The calculator sharpened conditionals; the guessing game mixed loops with logic; the shopping cart mixed dictionaries with math; file handling mixed lists, loops, and memory. Bit by bit, the pieces started talking to each other.</p><h3>Take-Home Assignment 📊</h3><p>My Week 1 submission is live on my GitHub account: <a href="https://github.com/Nneomadmapgirl">https://github.com/Nneomadmapgirl</a></p><p>It covers the basics I noted in this write-up: Python operations, simple problem-solving, and applying what we learned in a structured notebook.</p><p><strong><em>The Real Takeaway 👌👌👌👌</em></strong>: you don’t need to figure it all at once; it will get confusing, but little wins stack one on top of the other is a big encouragement. Check if a number is even. Build a calculator. Play with a list. Write a function. Save something to a file. Each tiny step builds the next one. In fact, understanding the Jupyter Notebook or Colab interface is a big step in learning. For me, one clear step at a time makes the difference, and that’s exactly how I plan to keep going. I’m not just learning tools; I’m learning how to make an impact.</p><p>See you all soon, from the dmapgirl</p><p>#DataScience #Machine Learning #AI #GIS #Cartography #Geology #Remote Sensing</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bd3f4b4b91e3" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>