A model of home values in my favorite neighborhoods in Oakland
Downloaded Redfin data for 7 elementary school districts in Oakland (Emerson, Piedmont, Rockridge, Crocker, Glenview, Chabot, Peralta), limited to a reasonable price range for single family houses (600k-1.5M) and within past 1.5 years.
Here are the result for the current model. See model.py for complete variable definitions. Dependent variable is price/sqft. Model includes:
- dummies per school district
- square feet (negative in price/sqft model)
- bathroom count
- 4 variables for different time periods
- whether I favorited the house
- lot size
- prewar (historic houses)
- walkscore
Bedroom count was not positive so was removed.
const 708.138537
SQUARE FEET -0.251568
BATHS 34.057259
recent 49.545652
target -57.288701
thisspring 9.578131
lastfall -39.910019
FAVORITE 58.796615
walkscore 1.929512
LOT SIZE 0.014656
historic 33.158759
chabot 156.026660
peralta 79.286202
crocker_highlands 77.134351
piedmont 56.810591
emerson 0.387973
To run again
- sample.py merges data and runs filters
- add_walkscore.py fetches walkscore for the whole sample
- model.py runs the model