documentation.txt

In it I explore the dataset, clean it and transform it. I then add more information to each property sample using census data from https://datashine.org.uk/.
I then train XGBoost to predict house price (with trend and seasonality removed) given census data and property type. 


Census data added:

sgC2: social grade C percentage
high depri: High deprivation percentage
low depri: Low deprivation percentage
ownership: home ownership percentage
bad_health: bad health percentage
second_add_uk: people with second address in the UK percentage
second_add_out: people with second address outside of the UK
train_to_work: percent of people taking train to work