Real State Business Analysis
Technology
Apache Airflow
Amazon Web Services
Scikit Learn
Python Pandas
Customer
Project Objectives
• Create a model in order to determine wheter property prices are undervalued or overvalued, with
all other variables being the same.
• Filter the under valued properties and create marketing campaigns around them using the
undervalued Price as a competitive advantage.
• Measure efectiveness of sales campaign around this strategy.
Machine learning operations (MLOps) steps
-
We used historical sales in Panama for the real state business with different characteristics (50 variables) and ultimate selling Price.
-
We also performed web scraping to obtain sales prices and characteristics on properties currently on the market.
-
We applied Cross-Validation on the data.
-
We used different Regression algorithms and some Decision trees algorithms for regression including XGBoost.
-
The best result we got from XGBoost: