- Simple linear regression
- Loss function: MSE, MAE
- Manual backpropagation
- Built and Trained the model
Dataset: Housing.csv
To predict the price of houses based on the given features.
- Performed Exploratory Data Analysis
- Handled the skewness and outliers from the features:
price
andarea
. - Used Log transformation to handle skewness.
- Analyzed categorical features and perform nominal encoding on them except for
furnishingstatus
. - For
furnishingstatus
, checked out ordinal as well as one-hot encoding to see what gives better result. - Analyzed Correlation plots and dropped some features having low correlation with the target. --
hotwaterheating
. - Checked if independent features are also corelated or not. Found some. But dropping them didn't improve the overall result.\
- Split the dataset into train and test set and also scale the data.
- Built Linear Regression Model.
- Used
MSE
,MAE
andLog(cosh(x))
as the cost function. - Trained the model on each one of them.
- Used R2 score as the evaluation metric.
Training Set : 0.6921
Test Set : 0.6837