tingly-amua/Under5-Mortality-Model

Predictive modeling of under-5 years mortality using the 2022 Kenya DHS data, combining descriptive statistics and machine learning to identify key determinants and support early intervention strategies. # Predictive Modeling of Under-5 Mortality Determinants in Kenya Using KDHS 2022 Data **Project Overview** This project focuses on analyzing and predicting under-5 mortality rates using demographic,health and social economic indicators. The workflow covers data cleaning,exploratory data analysis and modeling to indentify key factors associated with child mortality and build predictive models. **Repository structure** All files used in this project have been sorted and placed in folders specific to the tasks they carried out. These folders and their contents include: - Notebooks: Data cleaning.ipynb (data processing and feature engineering), EDA.ipynb-Exploratory (data analysis and visualization), Statistical_Associations.ipynb, Modelling.ipynb-Machine (learning models and evaluation) - Datasets: u5mr_subset.csv(first csv created after variable selection), u5mr_clean.csv(created after completion of data cleaning) - Deployment: Files pertinent to pickling and deployment, i.e. python-version, Procfile, app.py, model.pkl and requirements.txt files - Readme - Images: a collection of images generated and utilized in the project - presentation.pdf: a pdf presentation containing slides summarizing the objectives, highlights, and findings from modelling - Tableau visualization located here **Business understanding** *Objectives*: - Provide actionable insights on high-risk populations - Support targeted interventions (e.g., immunization, nutrition, maternal health services) - Guide equitable resource allocation - Ultimately reduce preventable child deaths *Stakeholders*: - policy makers - Public health agencies - NGOs and implementing partners - Academic researchers **Workflow** 1.Data Cleaning - Rename DHS codes such as V012, V013 to interpretable titles, ie. "Respondents current age", "Age in 5 year groups" etc - Aligned each variable with their datatype i.e categorical, int, or float - Map each column's code to their named counterparts eg. for "Sex of c …