In this repository one can find a working example of a generic Machine Learning project.
The idea is to offer a scaffolding for a general ML project which can be modified accordingly to one's own needs.
It can also be used as teaching material to Junior Data Scientists.
Namely, there are:
- full EDA (Exploratory Data Analysis) of the dataset;
- feature scaling, one-hot encoding, and feature engineering;
- analysis for selecting a suitable ML model;
- model training;
- hyperparameter tuning;
- model evaluation and computation of several relevant metrics;
- saving of the model as a serialized file;
- making predictions using the saved trained model.
Everything is as automated as possible: for instance, it is enough to insert the input data file and the target column name, and the notebook tries to automatically infer the correct ML task (binary classification, multiclass classification, regression).
Everything is described inside the notebooks/ml_project_template.ipynb notebook.
On a Windows machine, a solution for running a Makefile is also needed.
To install it, run
make setup
To simply use the application, run
make run
To transform notebooks in HTML reports without notebook execution, run
make reports_no_execute