Meet the Housing Data

HomeWorth has just bought a dataset. It covers more than twenty thousand districts in California, taken from a real census. It will be your training ground for the next few hours. Before any model touches it, a professional does one thing first. They look at it.

• Real census data
• Over twenty thousand districts
• First rule: look at the data

First look at the data

We load the data as a table, called a data frame. Shape tells us the size, twenty thousand six hundred and forty rows, and nine columns. The columns list tells us what each one means. And head shows the first few rows. Each row is a district. Med inc is the median income. House age, and average rooms, describe the homes. Latitude and longitude tell us where it is. And the final column, med house val, is the price, in units of a hundred thousand dollars. That is the number we want to predict.

from sklearn.datasets import fetch_california_housing

housing = fetch_california_housing(as_frame=True)
df = housing.frame

print(df.shape)
print(df.columns.tolist())
print(df.head(3).round(2).to_string())

(20640, 9)
['MedInc', 'HouseAge', 'AveRooms', 'AveBedrms', 'Population', 'AveOccup', 'Latitude', 'Longitude', 'MedHouseVal']
   MedInc  HouseAge  AveRooms  AveBedrms  Population  AveOccup  Latitude  Longitude  MedHouseVal
0    8.33      41.0      6.98       1.02       322.0      2.56     37.88    -122.23         4.53
1    8.30      21.0      6.24       0.97      2401.0      2.11     37.86    -122.22         3.58
2    7.26      52.0      8.29       1.07       496.0      2.80     37.85    -122.24         3.52

Features and target

Two words will now appear in every lesson. Features are the inputs, the facts we know about each example. By tradition, we call them capital X. The target is the output, the thing we want to predict, called small y. In our data, income, house age, and location are features, and the price is the target. Each row is one example. Fit learns the link from X to y.

Summary