I used gradient boosting regression and random forest regression to predict real estate prices for the kaggle competition titled "House Prices: Advanced Regression Techniques"

Data: The data for this analysis can be found at https://www.kaggle.com/c/house-prices-advanced-regression-techniques/data

There are 37 features in the dataset that describe physical attributes of each house.

Figure 1: Heat map showing the correlation among features

This dataset required a lot of cleaning due to the number of NAs:

I filled in NAs for features where it was logical to do so, and removed those that had more than 80% missing.

Additionally many features were skewed, including the response variable Price:

I used a box cox transformation on skewed features, and a log transformation on the response variable.

I applied both a gradient boosting regression model and a random forest regressor to predict price.

Hyperparameters were optimized using RandomGridSearch. For the gradient boosting regression model, I optimized:

I optimized the following hyperparameters for the random forest regressor:

The two models were compared given cross validation scores; the gradient boosting regressor had superior performance.

The gradient boosting regression model performed with a RMSE value of 0.1308 on the test set, not bad!

Additional explanations and code can be found at: https://towardsdatascience.com/home-value-prediction-2de1c293853c?source=friends_link&sk=704e52c7130e767aa32d6f6bae9678b9