Zillow Project - README

Table of Contents:

Project Description

Project Goals

Hypothesis/Questions

Data Dictionary

Planning

Instruction To Replicate

Takeaways

Project Description:

Back to 'Table of Contents'

Using the 'zillow' dataset from a SQL database acquire properties that have a transaction date of 2017 and are single family/single family inferred homes in order to best predict the home's value.

Project Goals:

Back to 'Table of Contents'

Implement Data Science Pipeline

Acquire 'zillow' Dataset

Prepare 'zillow' Dataset

Explore 'zillow' Dataset

Model 'zillow' Dataset

Deliver 'zillow' Dataset

Hypothesis/Questions:

Back to 'Table of Contents'

Hypothesis:

Given the 'zillow' dataset at hand, I believe that the location of the home, the contents of the home, as well as it's proximity to key features in the area will be a strong determining factor to accurately predict the home's value.

Questions:

Do values correlate to the location? (State, county, city, neighborhood)

Do values correlate with the amount of bedrooms and bathrooms

Do values correlate with the size of the home and it's overall property

Does the ratio of the home to overall sqft matter

Does proximity to key features like city center, leisure, stores, etc. matter

Does the local crime rate matter

Does the population density matter

Does proximity to job opportunity density matter

Does proximity to major roads matter

Data Dictionary:

Back to 'Table of Contents'

Field Name	Data Type	Data Format	Description	Example
	object	str	Defines gender of customer	'Male'
bedrooms	float	#.#	Defines the number of bedrooms in the home	3.0
home_sqft	float	#.#	Defines the total square footage of the home	2444.0
full_bathrooms	int	#	Defines the number of full bathrooms in the home	3
lotsize_sqft	float	#.#	Defines the total square footage of the lot the home resides on	10200.0
home_age	int	#	Defines how old the home is from when it was built to 2017	76
value	float	#.#	TARGET VALUE - Defines the home's value	689354.0
home_lot_ratio	float	#.#	Defines the ratio of the home size to the lot size	0.24
DUMMY COLS	uint	0, 1	Binary (True, False) column for specific column name	0

Planning:

Back to 'Table of Contents'

Planning

This good ol’ thingy

Acquire

env.file

SQL query

acquire.py

acquire.ipynb

Prepare

Remove unwanted cols

Aggregate cols

Confirm veracity of cols

prepare.py

prepare.ipynb

Explore

Determine best cols

Visualize regression lines to target variable

explore.py

explore.ipynb

Modeling

Linear Regression

LassoLars

TweedieRegressor

Polynomial Regression

Top Model

modeling.py

modeling.ipynb

Delivery

Final notebook (.ipynb)

final.py

Readme.md

Instructions To Replicate:

Back to 'Table of Contents'

1. Clone this repo
2. Create 'env.py' file that connects to SQL
3. Run desired .ipynb
   
   
   

Takeaways:

Back to 'Table of Contents'

Key Findings:

• Adds better predictive value to regression model
  • Number of bedrooms
  • Total sq. ft. of home
  • Number of full bathrooms
  • Total sq. ft. of lot the home is on
  • The age of the home
  • The ratio of the home and lot sq. ft.
    
    

Recommendations:

• Increases value
  • Number of bedrooms
  • Total sq. ft. of home
  • Number of full bathrooms
  • The ratio of the home and lot sq. ft.
• Decreases value
  • The age of the home
    
    

Takeaways:

• Focus for higher value
  • More bedrooms
  • More full bathrooms
  • Larger home
  • Larger home to lot ratio
    
    

Next Steps:

• Location specific
  • Difference in states
  • Difference in county
  • Difference in city
  • Difference in neighborhood
• Proximity specific
  • Density of schools
  • Density of entertainment/recreation
  • Density of landmarks/parks
  • Density of retail
  • Density of job opportunity
  • Accessibility
• Community specific
  • Density of population
  • Type of religion
  • Type of residents (Young, middle, old)
  • Type of family structures
  • Ethnic distribution
  • Gender distribution
• Hazard specific
  • Natural disaster risk (Tornado, flood, hurricane, etc.)
  • Crime rate density
  • Type of crime