The dataset is available here. Let's download it and store it in the data folder in the 01-KNN directory with the following commands:
cd ~/code/<user.github_nickname>/{{local_path_to("05-ML/03-Performance-metrics/01-KNN")}}
curl https://wagon-public-datasets.s3.amazonaws.com/Machine%20Learning%20Datasets/ML_Houses_clean.csv > data/houses_clean.csv- The dataset is a selection of features of the Houses dataset.
- Most features have already been preprocessed
- The target is the sale price of the houses
The exercise decomposes the KNN algorithm and its mechanics. You will:
- Discover its sensitivity to scale
- Voluntarily make it overfit
- Fine tune its parameter K
Finally, you will compare its performance to the one of a Linear Regression, and choose the model most suited for the task.
To start the exercise, open KNN.ipynb in jupyter notebook and follow the instructions.
🚀 Your turn!