K-Atina/Data-Sprint

A repository containing files related to the Strathmore/ I-Lab Africa Data Hackathon # DataSprint 2026 — Kenya FinAccess Financial Status Prediction **Strathmore Data Community × iLab Africa** --- ## Table of Contents 1. Project Overview 2. Dataset 3. Environment & Dependencies 4. Methodology - Data Loading & Inspection - Exploratory Data Analysis (EDA) - Preprocessing - Modelling - Evaluation - Feature Importance 5. Key Findings 6. Model Performance 7. Recommendations --- ## 1. Project Overview This project was built for **DataSprint 2026**, a one-week data science challenge organised by the Strathmore Data Community (SDC) in collaboration with iLab Africa. The challenge uses the **2024 FinAccess Household Survey** with the goal being to build a machine learning model that predicts whether a Kenyan adult's financial situation has **Improved**, **Stayed the same**, or **Worsened** compared to the previous year, and to surface the key drivers behind financial deterioration. --- ## 2. Dataset **Source:** Kaggle — Kenya FinAccess Household Survey 2024 **Official report & data manual:** finaccess.knbs.or.ke **File:** `finaccess2024_datasprint.xlsx` (also available as `.csv`) — **20,871 rows × 28 columns** ### Feature Categories | Category | Columns | |----------|---------| | **Demographics** | `county`, `location_type`, `Sex`, `Age`, `household_size`, `education_level`, `marital_status`, `has_disability` | | **Livelihood & Income** | `monthly_income`, `experienced_shock` | | **Mobile & Digital** | `mobile_ownership_1`, `mobile_money_access`, `barriers_mobile_money` | | **Financial Behaviour** | `Savings_formal`, `Savings_informal`, `Loan_formal`, `Loan_informal`, `defaulted`, `formal_service_use`, `barriers_bank`, `prodsum1` | | **Financial Health & Literacy** | `fl_score`, `nfhi_11`, `nfhi_12`, `nfhi_13`, `accessto_13k_1month`, `not_difficult` | | **Target** | `financial_status` | ### Target Class Distribution | Class | Count | Percentage | |-------|-------|-----------| | Worsened | ~10,978 | 52.6% | | Stayed the same | ~5,614 | 26.9% | | …