Project
Predicting House Prices with Machine Learning
February 9, 2022
- Investigated the dataset through descriptive and exploratory analysis
- Checked the distribution of the data and the correlation between features
- Investigated missing data and outliers
- Transformed the data towards a normal distribution and generated new features
- Trained a regression model explaining around 80% of the variance in sale prices
The heat map shows the features most strongly correlated with sale price. These were taken forward as the main features in the next steps.
After handling missing data and outliers, a regression model was trained. Plotting real sale prices against predicted prices shows the predictions are highly accurate: the closer the points are to the line, the better the prediction.