This is the final project for a Machine Learning lesson from Master 2 SISE at the Université Lumière Lyon 2. The aim of this project was to build an application using Dash to predict property values. Here, there is only a Jupyter Notebook with steps we followed to analyse the data. The app is available here (and more information regarding this project).
On this repository, there is only a Jupyter Notebook with steps we followed to analyse the data. The app is available here (and more information regarding this project).
We used about 3 million actual public data from data.gouv.fr, which are about property values from 2018, 2019, 2020 and 2021.
We first ran classification algorithms, including decision trees and k-Nearest Neighbors (with and/or without GridSearch), to predict the target "Type local" given relevant features. This first target then helped us to predict property values using regression algorithms such as linear regression, decision trees, and random forests.
Hugo ANDRE--ANTICHAN, Léo GONDOUIN, Annabelle NARSAMA, Arthur PARMENTIER