Materials for the practical sessions of the course "AI for business". Master in digital driven business, HvA.
This repository contains three folders:
- forecasting_case_study. This folder contains a (set of) jupyter notebooks to illustrate an end-to-end machine learning analysis for the problem of forecasting energy usage of households, as measured by smart meters. It illustrates the fundamental techniques to deal with forecasting problems as regression problems using tree models (xgboost) and neural networks. Note that the code in this folder is essentially just a refactoring into jupyter notebooks of the proper machine learning pipeline at https://github.com/riccardopinosio/DDB_smart_meters_case_study. Take a look at that repo too to see how an analysis from jupyter notebooks can be converted to a python package.
- credit_risk_case_study. This folder contains a (set of) jupyter notebooks to illustrate an end-to-end machine learning analysis for the problem of predicting default on credit card debt. It illustrates the fundamental techniques to deal with a classification problem using SVM.
- dataset. This folder contains the dataset for the credit risk case study, originally available at https://archive.ics.uci.edu/ml/datasets/default+of+credit+card+clients#. The dataset for the forecasting case study can be retrieved at https://www.kaggle.com/jeanmidev/smart-meters-in-london, as it is too large to include here.
For the material in the presentation, please see brightspace.