Just Enough Bend — medium Machine Learning in Code problem | Incognition

Just Enough Bend

10 pts · 10 coins

Machine Learning in CodeMedium

Just Enough Bend

More bend costs more columns, so you want the smallest degree that already fits the data well.

Write smallest_good_degree(xs, ys, max_degree, target).

Try each degree from 1 up to max_degree, in order.
For each one, fit make_pipeline(PolynomialFeatures(degree), LinearRegression()) on all the points, with xs as the only feature.
Score it on the same points with score (this gives R²).
Return the first degree whose R² is at least target, as an int.
If no degree up to max_degree reaches target, return 0.

Compare the unrounded R² with target.

Examples

xs = [0, 1, 2, 3, 4, 5, 6]

ys = [2.0, 1.1, 2.1, 4.9, 10.2, 16.8, 26.1]

max_degree = 4

target = 0.99

returns 2

xs = [1, 2, 3, 4, 5]

ys = [3.1, 4.9, 7.2, 8.8, 11.1]

max_degree = 3

target = 0.95

returns 1

Hints

solution.py

1import numpy as np
2from sklearn.linear_model import LinearRegression
3from sklearn.pipeline import make_pipeline
4from sklearn.preprocessing import PolynomialFeatures
5
6
7def smallest_good_degree(xs, ys, max_degree, target):
8    pass
9