← Back to Projects
House Price Prediction Using Regression
Predicting house sale prices using multiple regression models on 1,460 property records with 81 features. Ridge Regression achieved the best performance with R² = 0.8955.
🏠Property Features Input
📊Prediction Results
🏡
Enter property features and click predict to see the estimated sale price
🔬Model Comparison Results
| Model | R² Score | MAE ($) | RMSE ($) | MAPE |
|---|---|---|---|---|
| Ridge RegressionBest | 0.8955 | $2,942.56 | $3,684.29 | 1.49% |
| Bagging Regression | 0.0261 | $9,781.09 | $15,972.47 | 5.02% |
| Random Forest | -0.0056 | $10,507.12 | $16,230.35 | 5.41% |
| Lasso Regression | -1.9921 | $21,879.71 | $27,995.97 | 11.11% |
| Decision Tree | -1.3402 | $20,150 | $24,759.14 | 9.86% |
| PLS Regression | -0.5684 | $16,436.5 | $20,269.13 | 8.35% |
1,460
Property Records
81
Features Analyzed
6
Models Compared
Technical Implementation
Data Preprocessing
- Dropped features with more than 50% missing data
- Imputed numerical values with median, categorical with mode
- Applied label encoding for categorical variables
- Removed outliers using IQR method
- Standardized features for regularization models
Key Findings
- GrLivArea: Highest correlation with SalePrice (0.71)
- OverallQual: Strong predictor (0.79 correlation)
- Ridge Regression captures 90% variance in prices
- L2 regularization effectively handles multicollinearity
- Lasso selected 39 out of 74 features