This example uses the LinRegInternal and LinReg functions to find the coefficients of the variables that determine the selling price of a home in a given neighborhood. The response variable, SellingPrice (the selling price of the house in dollars) is modeled on these independent (predictor) variables:
- House size (in square feet)
- Lot size (in square feet)
- Number of bedrooms
- Whether the kitchen counter is granite (0 or 1)
- Whether the bathrooms are upgraded (0 or 1)
Input
| housesize | lotsize | bedrooms | granite | upgradedbathroom | sellingprice |
|---|---|---|---|---|---|
| 2397 | 14156 | 4 | 1 | 0 | 189900 |
| 2200 | 9600 | 4 | 0 | 1 | 195000 |
| 4032 | 10150 | 5 | 0 | 1 | 197900 |
| 3529 | 9191 | 6 | 0 | 0 | 205000 |
| 3247 | 10061 | 5 | 1 | 1 | 224900 |
| 2983 | 9365 | 5 | 0 | 1 | 230000 |
| 3536 | 19994 | 6 | 1 | 1 | 325000 |
| 3198 | 9669 | 5 | 1 | 1 |
The LinRegInternal function skips the last row of housing_data because it contains a NULL value.
SQL Call
SELECT * FROM LinReg (
ON LinRegInternal (
ON housing_data AS "input"
) AS "input" PARTITION BY 1
) AS dt;
Output
| coefficient_name | value_col |
|---|---|
| Intercept | -21739.2966650368 |
| housesize | -26.9307835091457 |
| lotsize | 6.33452410459345 |
| granite | 7140.67629349537 |
| upgradedbathroom | 43179.1998888263 |
| bedrooms | 44293.7605841832 |