papembengue0002-bit/healthconnect-experience-lab-data-science

Data Science track contribution to the AnalystLab Africa HealthConnect Experience Lab predicting patient appointment no-shows using appointment, booking and reminder data. # HealthConnect Experience Lab — Data Science Track (Week 4) Week 4 kickoff deliverables for the AnalystLab Africa Experience Lab shared project: **HealthConnect Clinic — Improving Patient Appointment Attendance and Healthcare Support Using Data and AI**. ## Project Context From Week 4 onward, all AnalystLab Africa internship tracks contribute to this shared project. This repository/folder contains the **Data Science track** contribution, focused on defining a machine learning problem to predict patient appointment no-shows. Week 4 is a foundation stage: understanding the problem, reviewing resources, and defining an approach — not building a final model. ## Contents | File | Description | |---|---| | `Week4_ML_Problem_Definition.ipynb` | Full notebook: data assessment, exploratory analysis, ML problem definition, target variable, candidate features, proposed modelling approach, key considerations | | `Week4_Project_Summary.docx` | Concise summary: problem addressed, resources used, key observations, proposed approach, key considerations, Week 5 focus | | `HealthConnect_Appointment_Data.csv` | Original project dataset (5,000 fictional, anonymised appointment records) | ## Key Findings - Dataset is structurally clean: no duplicates, no logical inconsistencies. - Missing `reminder_channel` values are structural (no reminder sent), not random — encoded as a category, not imputed. - **Prior no-show history** and **booking lead time** are the strongest candidate predictors identified so far (no-show rate ranges from 27.8% to 60.5% across lead-time buckets). - Proposed ML problem: binary classification (No-Show vs Attended), with Cancelled appointments (5.3%) excluded from the primary target. ## Tools Used Python, Pandas, NumPy, Matplotlib, Seaborn ## Author M'bengue mama — Data Science Intern, AnalystLab Africa #AnalystLabAfrica