WaweruGitau/Financial-Inclusion-Africa-

# Financial Inclusion in Africa — Bank Account Prediction A classification model predicting whether an individual is likely to have or use a bank account, built on the Zindi Financial Inclusion in Africa survey dataset covering respondents across Kenya, Rwanda, Tanzania, and Uganda. ## Approach **Exploratory data analysis** - Distribution and quartiles of respondent age and household size - Target variable (`bank_account`) class balance **Preprocessing** - One-hot encoding of categorical features (`relationship_with_head`, `marital_status`, `education_level`, `job_type`, `country`) - Correlation-based feature pruning — dropped features with near-zero correlation to the target - Min-max scaling of features into a `[0, 1]` range **Modeling** - Logistic Regression, trained on a stratified train/validation split - Evaluated with accuracy score and a confusion matrix - Final predictions generated on the competition test set and written out as a submission CSV ## Tech Stack - **Python** - **pandas** / **NumPy** — data wrangling - **scikit-learn** — preprocessing, Logistic Regression, evaluation metrics - **Seaborn** / **Matplotlib** — exploratory visualizations - **Jupyter Notebook** ## Running it 1. Install dependencies: ```bash pip install pandas numpy scikit-learn seaborn matplotlib jupyter ``` 2. Download the dataset from the Zindi competition page (`Train.csv`, `Test.csv`, `SampleSubmission.csv`, `VariableDefinitions.csv`) and update the working directory path in the notebook's second cell to point at your local copy. 3. Open `Financial Inclusion in Africa(Prediction).ipynb` in Jupyter and run all cells. ## Project Structure ``` . └── Financial Inclusion in Africa(Prediction).ipynb # EDA, preprocessing, model, submission ```