If you ask a prospective purchaser to define their ideal home, they usually will not start by talking about how high the basement ceiling is or how close it is to an east-west railroad. However, the goal of this study demonstrate that factors other than the number of beds or a white picket fence have a significant impact on price discussions.
I forecast the eventual price of each property using 79 explanatory factors that describe (nearly) every feature of residential dwellings in Iowa.
- Creative feature engineering
- Advanced regression techniques like random forest and gradient boosting
- Advanced preprocessing methodologies:
- Dealing with ordinal and nominal values [4 Methods]
- Dealing with missing values [8 Methods]
- Dealing with extraneous values 1 [Method]
- Finding the 'k' highly co-related 'SalePrice' attributes [1 Method]
- Removing duplicate values [1 Method]
- Illustrative visualizations
- In-depth testing with compared accuracy averages for every predictive model
View The Full Detailed Report Here --> LaTeX Report