Overview / Usage

Using XGBoost to check the feature importance on what customers would look for when searching for a house, the top three of importance were (Distance to station, House Age, and Number of Pubs). It was easier to get a better view of what customers were looking out for using by using two decision trees Gradient Boosting and XGBoost. A preprocessor was built to fit my data and transform my data.

Using Scikit-learn; the Simple imputer replaced missing values in my dataset, OneHot Encoder was used to transform, and the standard scaler. Also, The BaseEstimator and TransformerMixin set the parameters and fit the parameters respectively.

Methodology / Approach

Seaborn visualization tool was used to give a visual explanation of what values were missing and what axis needed to be dropped. Then a preprocessor was built using Scikit-learn to fit and transform my model. Then a prediction was made by comparing Ridge, Random Forest Regressor, Gradient Boosting Regressor, and XGBRegressor. However, Gradient Boosting Regressor came out to be the best prediction modeling tool using the R2 score and Mean Absolute Error calculation.

Finally, a graph for feature importance using XGBoost was built to view what customers looked at more when searching for a house. Then a Gradient Boosting and XGBoost decision tree was made to give a predicted target variable a customer would go for.

Technologies Used

Scikit-Learn/Sklearn

Numpy

XGBoost

Dill

Pandas

Numpy

Matplotlib

Seaborn

Repository

https://github.com/chelnnexy/Building-a-House-Price-Prediction