walternyamutamba-svg/IndabaX-Zimbabwe-Tutorial-2026

Tutorial from IndabaX Zimbabwe on applied data science. Covers a full industry workflow: exploratory analysis, feature engineering, model training and validation, and translating predictions into business decisions. Built on the Synthetic Bank Marketing dataset using Python and scikit-learn. # Data Science in Industry — IndabaX Zimbabwe Presenter: [Walter M Nyamutamba] Event: IndabaX Zimbabwe Date: [Thursday 4 March 2026] Contact: [walternyamutamba@aims.ac.rw] :[linkedin.com] ·Q[github.com] --- ## Overview This repository contains the materials from the IndabaX Zimbabwe tutorial session on applied data science. The session runs for 20 minutes of live coding followed by a 10-minute discussion. The central question throughout the tutorial is a real one: a bank runs outbound telephone campaigns to sell term deposit products. Given what we know about each customer, which ones should we call? The goal is not simply to build an accurate model — it is to show how a predictive model translates into a concrete operational decision that saves money and improves outcomes. The workflow follows the same sequence a working data scientist would use on the job. --- ## Repository Contents ``` . ├── IndabaX_Zimbabwe_Tutorial_PRESENTER.ipynb # Main tutorial notebook (presenter version) ├── README.md # This file └── datasets/ # Practice datasets for hands-on sessions ├── telco_churn.csv # Customer churn — binary classification ├── credit_risk.csv # Loan default prediction ├── hr_attrition.csv # Employee resignation risk ├── diabetes.csv # Patient health risk screening └── titanic.csv # Entry-level classification (well-documented) ``` --- ## What the Tutorial Covers **Section 1 — Environment setup and data loading** Imports, reproducible data loading, and an offline fallback in case the venue has no internet access. **Section 2 — Exploratory data analysis** Understanding the dataset before modelling: class imbalance, conversion rates by customer segment, and correlations among numerical features. **Section 3 — Data c …