Kaggle Competition (Getting Started): House Prices: Advanced Regression Techniques (Competition Here)
Predict sales prices and practice feature engineering, RFs, and gradient boosting
Software Used: Anaconda, Python 3.8
I have provided requirments.txt (or environment.yml) (if needed).
Description:
+----------------------------------------------------------------------------------------------------------------------------------------+
| .ipynb | Describe/Operation Performed |
+----------------+-----------------------------------------------------------------------------------------------------------------------+
| Notebook-I: | Read train and test csv and perform handling missing data. |
+----------------+-----------------------------------------------------------------------------------------------------------------------+
| Notebook-II: | EDA and Splitting train into train,cv,and test portion. |
+----------------+-----------------------------------------------------------------------------------------------------------------------+
| Notebook-III: | Training RandomForest Regression using all Features and hyperparameters |
+----------------+-----------------------------------------------------------------------------------------------------------------------+
| Notebook-IV: | Perform Testing stage for above trained model and submitted to Kaggle. |
+----------------+-----------------------------------------------------------------------------------------------------------------------+
| Notebook-V: | Perform and Training using Feature Selection with RandomForest Regression with best parameters from Notebnook-III |
+----------------+-----------------------------------------------------------------------------------------------------------------------+
| Notebook-VI: | Perform Testing stage for above trained model and submitted to Kaggle. |
+----------------+-----------------------------------------------------------------------------------------------------------------------+
| Notebook-VII: | Training GradientBoosting Regression using all Features and hyperparameters |
+----------------+-----------------------------------------------------------------------------------------------------------------------+
| Notebook-VIII: | Perform Testing stage for above trained model and submitted to Kaggle. |
+----------------+-----------------------------------------------------------------------------------------------------------------------+
| Notebook-IX: | Perform and Training using Feature Selection with GradientBoosting Regression with best parameters from Notebnook-VII |
+----------------+-----------------------------------------------------------------------------------------------------------------------+
| Notebook-X: | Perform Testing stage for above trained model and submitted to Kaggle. |
+----------------+-----------------------------------------------------------------------------------------------------------------------+
Result:
+---------------------------------------------------------------+
| Features | Model | Test Score (Kaggle Submission) |
+-------------------+----------+--------------------------------+
| All Features | RF Model | 0.19276 |
+-------------------+----------+--------------------------------+
| Feature Selection | RF Model | 0.18691 |
+-------------------+----------+--------------------------------+
| All Feature | GD Model | 0.20075 |
+-------------------+----------+--------------------------------+
| Feature Selection | GD Model | 0.15973 |
+-------------------+----------+--------------------------------+