This repository contains the data and code of group 2 of the course Design of Data-Driven Business Operations at Eindhoven University of Technology (TU/e) 2020/2021
To run the code, you can simply run the py files. To inspect the outputs, in editors such as VSCode and Pycharm, you can run the code sections between # %%, like cells of a Jupyter Notebook.
All data files, both the initial files and all intermediately created files, are present, so each file can be run in any order. Besides, some pickled variables are stored here. Note, running some files may take quite long.
The models folder contains the decision tree models and the columns that were used for these models. These are used in lp.py
In merge_data.py an initial data preparation is performed, providing the basis for the remainder of the analysis. constants.py contains some constant dictionaries and lists that are used in this and several other files.
The descriptives are created in descriptive.py
Both preprocessing and clustering for the behavioural analysis are performed in behavioural.py
The preprocessing for the predictive (and prescriptive) modeling is performed in predictive_preprocessing.py. Then, the experiments/model development are performed in predictive_modeling.py. helpers.py contains a function for running hyperopt, which is used in predictive_modeling.py
For the prescriptive analytics, two files are used. lpclasslibrary.py contains classes that are used in lp.py. You can run lp.py to run an optimization for an auction. At the top of this file, the id of the auction can be entered (which must be one of the auctions in the sample data)
- gurobipy >= 9.1.2
- hyperopt >= 0.2.5
- matplotlib >= 3.3.2
- numpy >= 1.19.5
- pandas >= 1.1.3
- scipy >= 1.5.4
- seaborn >= 0.11.1
- sklearn >= 0.0
- tqdm >= 4.56.2