In this project, the objective was to construct realistic models using regression and evaluate their performance and efficiency in predicting housing prices. The goal was to accurately estimate the value of real estate and uncover relevant factors that directly influence property prices. The study used two different models, multiple linear regression and k-Nearest Neighbours (kNN), to predict housing prices.
The results showed that the multiple linear regression model had the lowest Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE) at 151,751.64, 58,443,560,506.06, and 241,751.03 respectively. The model was considered a good model if the MAE and RMSE were low, and therefore, this model was deemed to be the best suitable for predicting housing prices.
Further analysis of the k-NN model revealed that the k=10 regression model had the lowest RMSE at 261,231.41. This confirmed that with 10 nearest neighbors, the model performed better.
In conclusion, this project successfully applied multiple linear regression and k-NN models to predict housing prices and evaluated their performance. The multiple linear regression model was found to be the best model for predicting housing prices based on its low MAE, MSE, and RMSE. This study provides valuable insights into the factors that influence housing prices and highlights the importance of carefully selecting the right model when making predictions in real estate.
- Python
- SciKitLearn
- Pandas
- Numpy
- Matplotlib
