Regression Analysis on House Prices

Looking at single family homes in Niskayuna, NY from $200k to $300k, I took 12 samples and ran a quadratic regression fit. The result seems to make sense:

To estimate the listing price, follow these steps:

start with $415,419,
add $12000 X # of bedrooms X # of bathrooms
minus $50,000 X # of bedrooms
minus $27,000 X # of bathrooms
minus $71 X # of days on the market

This means the bathrooms are really valuable. Given a 3 bedroom house, each bathroom you add increases the house value by $12k. Given a 5 bedroom house, each bathroom you add increases the house value by $36k. Conversely, having more bedrooms but fewer bathrooms hurt your house value.

Also notice that the list price goes down with time on the market.

The other model derived from Zillow's estimates give us a separate piece of information:
start with $142,000,
add $4,400 X # of bedrooms X # of bathrooms
add $101 X # of square feet of living space
minus $27,700 X # of bedrooms
minus $5000 X # of bathrooms

Here the trade-off between bathroom and bedroom is similar, bathroom is valuable. But we can see that each square feet of living space is $101. For the same square footage, the highest valued configurations would be bathrooms more than bedrooms, then bathrooms = bedrooms, then bathroom fewer than bedrooms.