Zillow House Appraisal

Project Description

Zillow is an innovative online database that people around the world use to see available properties and land for sale. Zillow provides people who visit their site with an estimate of how much a home will sell for. Data was gathered on multiple characteristics of the single family homes that were sold in 2017. The goal of this project is to use this data to make predictions/appraise homes that are currently on the market. Different features of homes were analyzed and determined to have the most influence on property value. Machine learning algorithms were applied to these features and an estimate of the property was given with the intent of allowing customers to buy and sell homes that are closest to market value.

Project Goal

• Discover features that influence single family homes appraisal.
• Use features to develop a machine learning model to predict the value of single family homes.
• This information can be used to improve reliability of zillow's home estimates and increase website traffic

Initial Thoughts

My initial hypothesis is that the square feet of a home, the number of bedrooms, and the number of bathrooms will have the greatest influence on property value.

The Plan

• Aquire data from Sequal Server

• Prepare data

  • Create Engineered columns from existing data
    • tax_value
    • bedrooms
    • bathrooms
    • sqft

• Explore data in search of drivers of tax value

  • Answer the following initial questions
    • What is a property's appraisal value?
    • What variables correlate with each other?
    • Can the number of bedrooms a property have help us know a property's appraisal value?
    • Can the number of bathrooms a property have help us know a property's appraisal value?
    • Does the area of a property help us predict tax_value?

• Develop a Model to accurately predict market value of homes

  • Use factors identified in explore to build predictive models of different types
  • Evaluate models on train and validate data
  • Select the best model based lowest root mean square deviation
  • Evaluate the best model on test data

• Draw conclusions

Data Dictionary

Feature	Definition
tax_value(target)	this is the properties appraisal value
bedrooms	this is the number of bedrooms on the property
bathrooms	this is the number of bathrooms on the property
sqft	This number represents the total area of a property in square feet.
lm_rmse	root mean squared deviation for the classic Linear Regression model
lassopoly	root mean squared deviation for the classic LASSO model
poly_rmse	root mean squared deviation for the classic Polynomial Regression model
baseline	root mean squared deviation using baseline
LASSO	Least Absolute Shrinkage and Selection Operator
LARS	Least Angle Regression

Steps to Reproduce

1. Clone this repo.
2. Acquire the data from sql database
3. Put the data in the file containing the cloned repo.
4. Run notebook.

Takeaways and Conclusions

• Square feet of home and tax value have a positive correlation
  • The influence appears moderate
• bathrooms and bedrooms have a positive correlation
  • correlation appears weak
• Polynomial model is the most accurate but is still not very reliable
• not enough features to create an accurate model that predicts tax_value

Recommendations

• add another feature like block location to come up with a better model
• Use polynomial regression to make predictions
• Consider finding another model or gathering more data