roma-glushko / kaggle-house-prices

🏘 Ames house dataset modelled and explained

Kaggle Competition: Ames House Sale Price

This repository contains explanatory data analysis and modeling of Ames house sale prices. Ames house database is sale history of almost 3000 houses that were sold in Ames, Iowa from 2008 to 2010. Each lot is represented by 80 features that describe all aspects of the house:

• Living area, floor areas, number of rooms, frontage, etc
• Neighborhood and Zoning
• Roof, masonry veneer, exterior properties
• Basement and fundament properties
• Air condition, heating, electrical properties
• Kitchen and bathroom properties
• Garage, porch, fireplace, fence, pool, wood deck properties

Competition Link: https://www.kaggle.com/c/house-prices-advanced-regression-techniques

Content

• index.ipynb - a domain-driven explanatory data analysis with a focus on feature meaning and value
• baseline-model.ipynb - a baseline modeling of house sale price using LinearRegression
• model.ipynb - house price modeling using XGBoost applied on various engineered features

Kaggle

• https://www.kaggle.com/glushko/house-prices-domain-driven-eda-part-i - EDA (Part I)
• https://www.kaggle.com/glushko/house-prices-regression-modelling-part-ii - Modelling (Part II)