Predictive Analytics
For a Tech Aggregator in Real Estate
- The client based out of UK is a Tech Aggregator in Real Estate with interest in Investment Banking too
- The underlying Client was evaluating buying a pool of Greece properties with help of our Client
- The Client wanted to Construct a mathematical model to predict the Price and Rent of the Properties in Greece
- The predictions were to be data driven using data from various public sources. Price and rentals were to be predicted for each property in the portfolio
- We trained data collected from public sources.
- We cleaned data, corrected for outliers, blanks and erroneous values. We binned certain variables like floor, build year etc and got small number of bins .
- We evaluated several models, performed feature transformations, analyzed relationships and statistical properties .
- G-Square selected the best performing model (outside training sample), built models for confidence interval and applied on final data
- We predicted price, rentals and confidence intervals for each property and overall portfolio .
