Project Name

House Price Prediction

Predict the significant variables for a housing company which can determine the price of houses. And How well those variables describe the price of a house?

Table of Contents

• Steps followed for prediction task.
• Observations
• Conclusions
• Subjective Questions

Steps followed

• Reading and Understanding the data
• Data Cleaning
• Impute null values with meaningful data given in data definition.
• Drop column with high correlation based on correlation matrix.
• Data Encoding - creating dummies for categorical.
• Splitting the data into training and testing sets.
• Log transformation of target variable.
• Data Scaling : fitting scaler on train and transform on test
• Running Recursive Feature Elimination wto get 50 variables.
• Building linear regrssion model with 50 features.
• Ridge - Finding best params and alpha for ridge using GridSearchCV.
• Metric and Coefficient evaluation
• Lasso - Finding best params and alpha for ridge using GridSearchCV.
• Metric and Coefficient evaluation
• Model Evaluation and Comparison of Linear, Ridge and Lasso.
• Determining important features based on the magnitude of coefficients

Observations

• 50 features taken from RFE
• Linear regression model was overfitting the data because dataset is too small.
• Ridge and Lasso removes the overfitting and perfoems better

Conslusion

Important predictors for housing price were - GrLivArea, OverallQual, OverallCond, GarageCars, TotalBsmtSF, BsmtFinSF1, LotArea