<?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 Gbubemi Erics on Medium]]></title>
        <description><![CDATA[Stories by Gbubemi Erics on Medium]]></description>
        <link>https://medium.com/@gbubemierics?source=rss-89dc0c46c9c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*LSoO9y0xM8lR4Qz1</url>
            <title>Stories by Gbubemi Erics on Medium</title>
            <link>https://medium.com/@gbubemierics?source=rss-89dc0c46c9c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 10:17:59 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@gbubemierics/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[Green Journeys: Transforming Travel to Save Our Planet]]></title>
            <link>https://medium.com/@gbubemierics/green-journeys-transforming-travel-to-save-our-planet-c271b594036f?source=rss-89dc0c46c9c------2</link>
            <guid isPermaLink="false">https://medium.com/p/c271b594036f</guid>
            <dc:creator><![CDATA[Gbubemi Erics]]></dc:creator>
            <pubDate>Mon, 17 Mar 2025 08:17:15 GMT</pubDate>
            <atom:updated>2025-03-17T08:17:15.790Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Introduction</strong></p><p>Every year, travelers across the UK collectively cover billions of miles for work, leisure and events. While travel remains an essential part of life, it also poses an increasing challenge to the environment. In 2021, transportation accounted for 26% of the UK’s total greenhouse gas emissions, making it the largest emitting sector, surpassing energy and industry.</p><p>As the UK works toward achieving net-zero emissions by 2050, initiatives are being launched to explore sustainable travel solutions. This report aims to address three key questions:</p><ol><li>What is the greenest way to travel?</li><li>How does age influence travel choices?</li><li>When is the optimal time to travel to reduce emissions?</li></ol><p>By answering these questions, we can uncover actionable insights that balance convenience, cost and sustainability.</p><p><strong>Sustainable Travel: Cost vs. Emissions</strong></p><p>One of the most significant findings from the analysis is the relationship between travel costs and carbon emissions across various modes of transport. By visualizing the data, we can see clear trends in both environmental and economic impact.</p><p>Key findings include:</p><ul><li><strong>Walking is the greenest option</strong>, generating zero carbon emissions while being cost-free, making it the best choice for short distances.</li><li><strong>Trains, buses, and electric car-sharing options</strong> emerge as highly sustainable, producing significantly lower emissions per traveler compared to petrol cars.</li><li><strong>Electric car-sharing has relatively low cost and low emissions</strong>, presenting a promising solution that requires greater infrastructure investment to compete with public transport.</li><li><strong>Solo petrol car travel has the highest emissions and costs per mile</strong>, nearly five times higher than public transport per traveler.</li></ul><p>The general takeaway? The more we spend on private travel, the greater our carbon footprint tends to be.</p><p><strong>How Age Influences Travel Choices</strong></p><p>Analyzing travel preferences across different age groups provides valuable insights into transportation habits and how they evolve over time.</p><p><strong>Key Observations:</strong></p><ul><li>Young adults (18–24) show a clear preference for public transport, particularly buses and trains. Their budget-conscious behavior aligns well with sustainability.</li><li>The 25–34 age group still favors public transport but shows an increasing shift towards private vehicles, both petrol and electric, likely due to convenience.</li><li>Travelers aged 35+ rely heavily on private vehicles, especially petrol cars, contributing to higher emissions per person.</li></ul><p>The decline in public transport use as age increases suggests a need for targeted interventions that promote sustainable travel options for older demographics. Many older travelers prioritize comfort and convenience over sustainability, making mindset shifts a key challenge.</p><p><strong>Optimal Travel Times and Carpooling</strong></p><p>The timing of travel plays a significant role in determining emissions. Emissions vary depending on whether travel occurs during peak or off-peak hours, on weekdays or weekends.</p><p><strong>Key Insights:</strong></p><ul><li><strong>Weekday off-peak travel produces up to 27% lower emissions</strong> than peak-hour travel, due to reduced congestion and smoother traffic flow.</li><li><strong>Weekend peak-hour travel has the highest emissions</strong>, with congestion and stop-and-go traffic increasing fuel consumption significantly.</li><li><strong>Carpooling can reduce per-person emissions by up to 40%</strong>, making it a key strategy for sustainable travel.</li></ul><p>By optimizing travel times and encouraging carpooling, emissions can be significantly reduced without sacrificing convenience.</p><p><strong>Recommendations for Sustainable Travel</strong></p><p>To minimize environmental impact and promote sustainability, the following strategies are recommended:</p><ol><li><strong>Encourage the use of trains and buses</strong>, which emit far less CO₂ than private vehicles, and promote walking for short distances.</li><li><strong>Support carpooling initiatives</strong> by offering parking incentives for shared rides, reducing emissions per traveler by up to 40%.</li><li><strong>Promote off-peak travel</strong>, particularly on weekdays, to reduce congestion and lower emissions by as much as 30%.</li><li><strong>Tailor sustainability campaigns for older travelers</strong>, emphasizing the financial savings and environmental benefits of public transport and carpooling.</li><li><strong>Improve weekend public transport options</strong>, ensuring accessibility for all demographics and reducing dependency on private vehicles.</li></ol><p><strong>Conclusion</strong></p><p>Transportation is a major contributor to greenhouse gas emissions, but it also presents an opportunity for meaningful change. By adopting sustainable travel habits, individuals, policymakers, and organizations can collectively reduce emissions, ease congestion, and foster a greener travel culture.</p><p>Simple shifts in behavior — such as traveling off-peak, walking shorter distances, or sharing rides — can deliver substantial environmental and financial benefits. As the UK moves toward its net-zero emissions targets, these insights offer a roadmap for a more sustainable future.</p><p>Sustainable travel isn’t about restricting movement — it’s about making choices that enrich both our lives and the planet.</p><p><strong>Sources</strong></p><ul><li>Burnett, N., Stewart, I., Hinson, S., Tyers, R., Hutton, G., &amp; Malik, X. (2024). The UK’s plans and progress to reach net zero by 2050. House of Commons Library. <a href="https://commonslibrary.parliament.uk/research-briefings/cbp-9888/">Link</a></li><li>National Highways (2023). Connecting the Country: Our Long-Term Strategic Plan to 2050. <a href="https://nationalhighways.co.uk/media/0hpe3wh2/cre22_0150-masterplan-national-highways-ris3_final-1.pdf">Link</a></li><li>Egbumokei, C., Christenson, A., &amp; Barrand, S. (2023). UK Environmental Accounts — Office for National Statistics. <a href="https://www.ons.gov.uk/economy/environmentalaccounts/bulletins/ukenvironmentalaccounts/2023">Link</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c271b594036f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My First Experience with Data Science: Predicting Divorce Using Machine Learning]]></title>
            <link>https://medium.com/@gbubemierics/my-first-experience-with-data-science-predicting-divorce-using-machine-learning-bdcb721ae74c?source=rss-89dc0c46c9c------2</link>
            <guid isPermaLink="false">https://medium.com/p/bdcb721ae74c</guid>
            <dc:creator><![CDATA[Gbubemi Erics]]></dc:creator>
            <pubDate>Wed, 05 Feb 2025 08:02:41 GMT</pubDate>
            <atom:updated>2025-02-05T08:02:41.338Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RPjlqd_8f9HTCOHi-nwliQ.jpeg" /></figure><p><strong>My First Experience with Data Science: Predicting Divorce Using Machine Learning</strong></p><p><strong>Introduction</strong></p><p>In this article, I will walk you through my first personal data science project, which aims to classify couples who are likely to divorce or remain happily married based on their responses to a structured questionnaire. This project not only strengthened my technical skills but also demonstrated the real-world application of machine learning in understanding human behavior. The article includes the projects code snippets, ensuring reproducibility for aspiring data scientists and professionals alike. You can find the complete code on my <a href="https://github.com/gbubemierics/Divorce_Prediction">GitHub repository (http://bit.ly/4jwEEap) and access the dataset from</a> the <a href="https://archive.ics.uci.edu/dataset/539/divorce+predictors+data+set">UCI Machine Learning Repository (https://bit.ly/4jDhbEw)</a></p><p><strong>Aim</strong></p><p>The objective of this project is to predict couples based on their likelihood of divorce or remaining happily married using a dataset containing relationship-related questions.</p><p><strong>Dataset</strong></p><p>The dataset originates from Turkey and is publicly available in the <a href="https://archive.ics.uci.edu/dataset/539/divorce+predictors+data+set">UCI Machine Learning Repository</a>. It includes responses from 170 couples (either divorced or happily married) to 54 questions. Each response is rated on a 5-point Likert scale:</p><ul><li>0: Never 1: Seldom 2: Average 3: Frequently 4: Always</li></ul><p>This scale allows us to quantitatively measure relationship dynamics, making it suitable for predictive modeling.</p><p><strong>Project Workflow</strong></p><p>The project follows a systematic approach divided into several steps:</p><p><strong>Step 1: Import Libraries and Load Dataset</strong></p><p>To start, we imported the necessary Python libraries and loaded the dataset:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*4JSdPobOicpZ1PamB6eP_A.png" /><figcaption>Importing the libraries for this project.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*r8cN_5Iga8g3BPL4eVDBeA.png" /></figure><p>The above snippet shows the overview of the data</p><p><strong>Step 2: Exploratory Data Analysis (EDA)</strong></p><p>Conducting EDA is crucial for understanding the dataset’s structure and identifying any patterns or anomalies. The EDA revealed no missing values, all variables are integers and the class distribution was relatively balanced as shown in the image below, which ensures unbiased model training.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*DCN3bLQcBExwuSRTmrc5_A.png" /></figure><p>The above visualization shows a 50–50 distribution between divorce and not divorce couple.</p><p><strong>Step 3: Feature Selection</strong></p><p>Feature selection is a crucial step in improving model performance by identifying the most relevant predictors while reducing noise. The dataset contains 54 relationship-related questions, and not all features contribute equally to predicting divorce. To address this, the following techniques were applied:</p><ul><li>Correlation Analysis: Features highly correlated with the target variable were prioritized, while highly collinear features were removed to prevent redundancy.</li><li>Feature Importance Scores: Using algorithms like Random Forest, feature importance was computed to rank the predictive power of each variable.</li><li>Recursive Feature Elimination (RFE): This method iteratively removes the least significant features to enhance model efficiency and generalization.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*pHgupgrAFEnH1x4h0yPSCA.png" /></figure><p>By selecting the top features as shown in the graph below, we optimized computational efficiency while maintaining high predictive accuracy.</p><p>The dataset is split into training and testing sets using an 80:20 ratio:</p><p><strong>Random Forest</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*z100ANvxTydYsI1Wov6jcw.png" /></figure><p><strong>Step 4: Data Splitting</strong></p><p>The dataset is split into training and testing sets using an 80:20 ratio:</p><p>X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)</p><p><strong>Step 5: Model Training</strong></p><p>Two models were trained and evaluated:</p><ul><li>K-Nearest Neighbors (KNN): Simple and effective for smaller datasets</li><li>Neural Network (MLPClassifier): capable of capturing complex relationships in the data</li></ul><p># KNN Classifier</p><p><strong>K-Nearest Neighbors (KNN)</strong> is a simple, instance-based learning algorithm that classifies data points based on the majority class of their closest neighbors. It works well for small datasets but can struggle with high-dimensional data and computational efficiency as the dataset grows.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*x4PZ7qSSgFvnOnjWTaW4pg.png" /></figure><p>Evaluation</p><p>The models were evaluated using accuracy, precision, recall, F1-score, and AUC-Score. A detailed classification report was generated for each model:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*0oz-qlW6VqHfl4s6D56gqA.png" /></figure><p><strong>Neural Network</strong></p><p><strong>Neural Networks (MLPClassifier) </strong>are powerful models that learn complex patterns through multiple layers of neurons. They excel at capturing nonlinear relationships in data, making them more effective for intricate datasets like relationship prediction, as seen in this project below</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*CHp_XjxDYsiF4H4qUUH-xg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/975/1*DrySsVDHx4TKcQhXbFq5lw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/965/1*jKALS8Ai442gcoYjEGqqyw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/884/1*PFmuT0pepkMqixh6MI-3Lw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/956/1*jH3vWQwvZHTjo6SHfzKgHg.png" /></figure><p><strong>Step 6: Evaluation</strong></p><p>The models were evaluated using accuracy, precision, recall, F1-score, and AUC-Score:</p><p><strong>Results</strong></p><ul><li><strong>KNN Classifier</strong>: Achieved an accuracy of 97%.</li><li><strong>Neural Network</strong>: Outperformed KNN with an accuracy of 99%.</li></ul><p>The Neural Network’s superior performance was attributed to its ability to capture complex relationships in the data.</p><p><strong>Further Analysis</strong></p><ol><li>Hyperparameter Tuning: Both models underwent hyperparameter optimisation to further improve accuracy.</li><li>Exploring Additional Models: Testing other classifiers, such as Support Vector Machines or Gradient Boosting, could enhance results.</li></ol><p><strong>Limitations</strong>: The dataset’s relatively small sample size and cultural specificity might limit its generalisability. Future research could involve larger, more diverse datasets.</p><p>As this was an early project here are 4 major changes I would have done differently if I had to do it over again</p><p>1. Exploratory Data Analysis (EDA) Deep Dive — Include more visualizations like pair plots and box plots to uncover hidden patterns.</p><p>2. Baseline Model Comparison — Use a simple logistic regression or a majority class predictor as a benchmark.</p><p>3. SHAP or Feature Importance Analysis — Explain model predictions by visualizing which features impact the outcome most.</p><p>4. Hyperparameter Optimisation for Neural Network — Fine-tune the architecture (e.g., number of layers, activation functions) for better performance.</p><p><strong>Conclusion</strong></p><p>This project demonstrates the power of machine learning in predicting relationship outcomes by analysing structured questionnaire responses. Through systematic data processing, model evaluation and feature selection, we achieved high accuracy in predicting divorce likelihood. Enhancements such as deeper exploratory data analysis, baseline model comparison and explainability techniques could further improve insights and model interpretability. Future work may involve larger, more diverse datasets and ethical considerations in predictive analytics. This study serves as a foundation for data-driven approaches in understanding human relationships and decision-making.</p><p>If you’d like to explore the code or try running this project yourself, you can access my full implementation on GitHub( http:<a href="https://github.com/gbubemierics/Divorce_Prediction/blob/main/README.md">/</a>/bit.ly/4jwEEap). The dataset is publicly available at the <a href="https://archive.ics.uci.edu/dataset/539/divorce+predictors+data+set">UCI Machine Learning Repository</a> (<a href="https://bit.ly/4jDhbEw">https://bit.ly/4jDhbEw</a>).</p><p>Feel free to contribute, ask questions, or share feedback!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bdcb721ae74c" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>