Our client is a Real Estate aggregator company that lists properties across the country on their platform. Property owners can enlist their properties in the platform, and the customers can directly contact the owners if they like to enquire about a property. The customers found that similar properties in the same area were significantly different in price. They have contacted the support team and raised the issue multiple times. This inconsistency in pricing is creating a lack of trust on the platform and hence the company called us to build a price discovery and regulation model that would estimate the price range of property given its attributes like area, apartment type, amenities, etc. This project involves building a regression model for price prediction, developing a web application for the same using the FAST API framework, and deploying it on Heroku.
To predict the price range of a new listed property based on attributes like area, apartment type, amenities, etc. Data Description The dataset contains information about 200 properties in Pune, Maharashtra, India, on various attributes such as area, amenities, description, apartment type, etc. Tech Stack
➔ Language: Python
➔ Libraries: pandas, numpy, scipy, matplotlib, seaborn, sklearn, nltk, statsmodel
Approach
- Data Reading
- Data Preprocessing
● Categorical Data Cleaning
● Continuous Data Cleaning
● Using Regex Library
● Univariate Data Analysis
● Multivariate Data Analysis
● Outlier Treatment
● Feature Extraction
● Text Data Processing
● Parts of Speech Tagging
● Count Vectorization and N-grams - ML Model Building
● Linear Regression
● Confidence Interval
● Regularization
● Ridge Regression
● Lasso Regression
● Voting Regressor - Model Deployment
● APIs
● Web Application Development using Streamlit
● Heroku Deployment
● Model Inference Pipeline - Testing
● A/B
● AA/B