I have tried to make the model as small as possible, at the same time have a score in the acceptable interval.

Data Tidying:

• Removing outliers did not have a positive effect, so I deleted that part to make the model as simple as possible

• Merged both train and test data to be able to find more reliable values for missing values.

• After finding the not-numerical columns, analyzing the reasoning behind being missing, decided to fill them with “None”

• For numerical columns, filled them with mean of the column.

Feature Engineering:

• Created a variable for some features that had correlation (had multi-collinearity)

• Applied log transform on the features having skewness over 0.2 and also on SalePrice (y).

• Using dummy variables to get rid of the categorical data.

• Split data again to test and train set accordingly.

Model Selection:

• Using LASSO model, with alpha 0.0005, based on trial and error, with default 3 fold cross validation.

Kaggle Score = 0.11939