In this project, we use machine learning algorithms to predict Bangalore house prices.
The dataset is sourced from Kaggle. Here's an overview of the steps we followed:
- Importing libraries such as pandas, numpy, and matplotlib.
- Importing the data into a DataFrame and examining its shape and data types.
- Dropping unimportant columns and handling any null values, which were found to be minimal.
- Performing feature engineering to create new, more interpretable features.
- Reducing the dimension of the data by focusing on the most frequent values.
- Removing outliers with the assumption that the data is normally distributed.
Our model building process can be summarized as follows:
- Splitting the data into the train and test sets.
- Utilizing GridSearchCV to determine the best model parameters for linear regression, Lasso regression, and decision tree.
- Training the model with the optimal parameters obtained.
- Evaluating the model's accuracy using K-fold cross-validation, resulting in an impressive accuracy score of 85.7%.