- This notebook is for people who are looking to buy a place in Bangalore city(INDIA)
- Created a model that predicts Bangalore house rate to help people to know about the prices of house in various places without the need of contacting different agents for the same.
- Deployment using streamlit on Heroku.
- Data was collected from Kaggle
- Data shape is 13320 rows and 9 columns.
-
Column 'total_sqft' was provided in avergare terms i.e 1195 - 1440, converted the same to float by taking the average of the numbers.
-
Created a new column - 'price_per_sqft' for better analysis.
-
Scaled down the number of locations to 241 from 1287 since most of the location occured less than 10 times.
-
On an average, the ratio between total_sqrt_foot and number of BHK should always be 300, hence we removed all the entries with values lesser than 300.
-
-
Here we find that min price per sqft is 267 rs/sqft whereas max is 12000000, this shows a wide variation in property prices. Removal of 2214 outliers using Mean and Standard Deviation
-
.png)
-
We see that there are alot of houses with 5000/- price per square feet rate
-
Removal of Data where the number of bathrooms is higher than number of bhk+2
.png)
- Performed One hot encoding to represent the categorical values in binary form since machine learning algorithms cannot operate on label data directly.
- I also split the data into train and tests sets with a test size of 20%. Model Used:-
- Multiple Linear Regression - r^2 value of 0.86
Built a function to predict the house price with location, number of Square foot area, Bathroom, and BHK.
The prices mentioned are in Lakhs(Indian Currency)
In this step, I built a streamlit API endpoint that was hosted on Heroku. The API endpoint takes in a request with a list of values from location, number of Square foot area, Bathroom, and BHK. Web_Page