Supervised machine learning

In this project, I will use the famous housing price prediction data and employ the two supervised ML algorithms (classification and regression). The dataset contains a total of 80 features. In doing so, I will apply many useful techniques (hyperparameter tuning, scikit-learn pipelines) to find the best-performing model. The best model will be then used to test the unlabeled data provided for an internally organized competition. For more details on the data used, click here

Classification: Predict whether the house is expensive (or not)

For this part of the project, I will predict the binary label "Expensive", which takes the value of 1 if the price of the house is above a certain threshold, e.g., $200,000, and 0 otherwise. I tested numerous state-of-the-art models, e.g., XGBoostClassifier() DecisionTreeClassifier(), RandomForestClassifier(), KNeighborsClassifier(), "LogisticRegression()*, and SVC().

The RandomForestClassifier() performed the best with the training (accuracy = 95.5%) and test data (accuracy = 95.9). The test accuracy of the best model on the unlabeled competition data was 97.3%.

A complete code is available in this repository click here.

Regression: Predict the exact price of the house

For this part of the project, I will use the continuous house price column: SalePrice. The prediction will take place by applying many regression estimators to get the best prediction performance, i.e., higher R2 squared and lower Root Mean Squared Error (RMSE) scores. Ultimately, the trained model will be used to predict the house prices of the unlabelled test dataset present on Kaggle (link here).

Complete code is available in this repository click here.