Ruth-Nafula/financial-status-kenya-prediction

##### **Financial Status Prediction in Kenya Using Machine Learning** ###### **Project Overview** This project builds a supervised machine learning model to predict the financial status of Kenyan adults. The goal is to classify whether someone's financial situation has: * Improved * Stayed the same * Worsened This is based on different factors such as demographics, livelihood and financial behavior. The main aim is not just prediction, but also understanding what factors are most linked to financial outcomes. ###### **Dataset Description and Preparation.** The dataset contains responses from 20,871 Kenyan adults collected through a structured survey covering socio-economic, demographic, and behavioural indicators. Before analysis, the dataset required cleaning and preprocessing. Duplicate records were removed to ensure data quality and prevent bias from repeated entries. Categorical variables were encoded to make them suitable for machine learning algorithms. Some features were also scaled to ensure consistency across categories. Following preprocessing, the dataset was split into training and testing sets in preparation for modelling. ###### **Exploratory Data Analysis(EDA)** Exploratory analysis was carried out to understand relationships between key variables and the target outcome. A key focus was the relationship between financial shocks and financial status, as well as how education level influences financial outcomes. These relationships were examined using visualisations such as grouped bar charts and heatmaps to highlight differences in proportions across categories. The EDA revealed clear patterns suggesting that certain socio-economic factors are strongly associated with financial outcomes, which informed feature expectations in the modelling stage. ###### **Modeling Approach** The modelling process was carried out as a comparison between two different classification approaches to evaluate performance differences between a n …