Video Presentation link: https://youtu.be/GKEGwVf87Rs
Choice 1: Regression on Ames Housing Dataset
You can load the Ames housing dataset from here. You can find a description of the variables here. Take note that for categorical variables, NA here does not mean a missing value, but should be treated as a separate category.
- Visualize the univariate distribution of each continuous input variable, and the distribution of the target variable. Write down anything you find worth noting. Is there something that might require special treatment?
- Visualize the dependency of the target on each continuous feature via scatter plot.
- Split data in training and test set. Do not use the test-set unless for a final evaluation in Step 6. Convert each categorical variable into numerical variables using one-hot-encoding. Example of one-hot encoding: Gender: Male -> (1, 0), Female -> (0, 1) Ethnicity: 1. Caucasian, 2. African American, 3. Hispanic, 4. Asian, 5 Native American, 6 Pacific Islander One-hot encoded labels: 1 → (1, 0, 0, 0, 0, 0) 2 → (0, 1, 0, 0, 0, 0) 6 → (0, 0, 0, 0, 0, 1)
- Evaluate Linear Regression, Polynomial Regression, Decision Trees, and Neural Network. Note that you need to decide the choice of hyper-parameters for the models, such as the degree of polynomial regression, the maximum depth of random forests, and the number of layers for neural networks.
- Based on your models, what are the most important factors for deciding the price of houses?