Machine Learning: University of Genoa (90948) - Fall 2019 - Final project
Live Notebook on Colab
Abstract
We present a pair-project in which we tackle the same problem proposed by the IREN company during C1A0 Hackathon. We tried to mix project of types 1 and 2 with the aim of considering a full real-case scenario concerning the electricity market. As for the type 2 part of the project, we provide a detailed data processing and a simple time-series analysis in the notebook file Code/Data processing/Time-series analysis.ipynb. As for the type 1 part of the project, a detailed formulation of two different algorithms follows: Random forests (by Gianvito Losapio) and Gradient boosting (by Federico Minutoli). Both of the algorithms have been implemented from scratch. The structure of the code is presented in the notebook file Code/Problem.ipynb as well as its use to solve the main problem. Results and comparisons of the methods are provided.