kristaaaaaaaaa/internet-affordability-zimbabwe

Internet Affordability in Zimbabwe – Machine Learning Analysis 📌 Project Overview This project explores internet affordability in Zimbabwe using data-driven analysis and machine learning models. It investigates how geographic, economic, and demographic factors influence affordability perceptions and internet access. 🎯 Objectives Analyze internet affordability using survey and performance data Identify key predictors of affordability perception Compare performance of machine learning models (Random Forest, etc.) Provide insights for policymakers, telecoms, and regulators 📂 Repository Contents Internet_Affordability_Analysis.Rmd – Main R Markdown analysis script Internet_Affordability_Analysis.html – Rendered HTML report internet_data.csv – Source dataset internet_affordability_model.rds – Saved Random Forest model preprocessing_recipe.rds – Preprocessing pipeline Predicting Internet Affordability Perception in Zimbabwe.pdf – Full project report 🔑 Key Findings App performance emerged as the strongest predictor of affordability perception Urban-income interactions showed significant importance Digital engagement scores correlate with affordability perception The Random Forest model achieved high accuracy and AUC 💡 Business & Policy Implications Improving technical performance may have stronger impact than reducing prices Geographic/economic targeting can optimize affordability strategies Engagement programs could improve digital inclusion 🚀 Deployment Readiness Model size: ~2 MB, suitable for real-time applications Preprocessing pipeline ensures consistent handling of new data Trained model generalizes well across demographic groups 🔮 Future Work Expand dataset for stronger generalization Include additional economic indicators Implement real-time monitoring and retraining pipelines 🛠️ Tech Stack Language: R Libraries: tidymodels, randomForest, ggplot2, dplyr Environment: RStudio, GitHub