Machine Learning Workflow: Regression use case

Project: Amsterdam House Price Prediction

Photo by Adrien Olichon on Unsplash

Objective: Predict House Pricing and Illustrate Machine Learning Workflow

Framework:

• The problem is framed as a regression problem since the target has continuous numerical values.
• I used a supervised learning approach
• The model will be trained offline in one batch
• The primary performance metric will be r2, combined with Root Mean Square Error (RSME) and Maximum Absolute Error (MAE)
• We aim for a model with an r2 of at least 0.7.

How would we solve the problem manually:

• One can derive the price of a given house by comparing the price with other similar houses, e.g., on pararius.nl.

List assumptions coming from research questions:

• the bigger the area of a house, the higher the cost of a house
• the bigger the number of rooms, the higher the cost of a house
• a closer location to the city centre would lead to higher housing costs.