Several Features

Part of the Supervised Learning section of Coddy's Machine Learning journey. Lesson 12 of 58.

A house's price depends on more than size: a long walk to the station lowers it. Learning from both, a model draws parallel lines:

Ten houses as dots, price against area: houses near the station in blue, far ones as orange squares. Two parallel model lines, for a 5 and a 25 minute walk; the 25 minute line sits 102 thousand dollars lower at every size

In code, X gets both columns and coef_ one weight per feature:

X = houses[["area_m2", "minutes_walk"]]
model = LinearRegression()
model.fit(X, houses["price"])
print(np.round(model.coef_, 1))

This prints:

[ 2.6 -5.1]

The weights follow X's columns. Each extra m² adds 2.6 thousand dollars; each extra minute's walk takes 5.1 thousand off, the other feature held still.

A weight's unit is the target's per unit of its feature: thousands per m², per minute.

Challenge

Medium

A caterer quotes new events from 40 past ones in events.csv: guests, courses, months_ahead (how early it was booked) and cost in dollars. The starter reads a new event into new_event.

Fit a LinearRegression of the cost on the features.
For each feature, print what one more adds or takes off (2 decimals).
Print the new event's quote (2 decimals).

For another caterer:

Each extra guest adds 41.5 dollars
Each extra course adds 198.25 dollars
Each extra month booked ahead takes off 22.8 dollars
Quote: 4312.6 dollars

Try it yourself

import numpy as np
import pandas as pd
from sklearn.linear_model import LinearRegression

# One row per past event: guests, courses, months booked ahead, cost in dollars
events = pd.read_csv("events.csv")
features = ["guests", "courses", "months_ahead"]
units = ["guest", "course", "month booked ahead"]
# The new event, as a table with one row
guests = int(input())
courses = int(input())
months = int(input())
new_event = pd.DataFrame({"guests": [guests], "courses": [courses], "months_ahead": [months]})

# Write your code below