How Much Each Direction Holds — easy Machine Learning in Code problem | Incognition

How Much Each Direction Holds

10 pts · 10 coins

Machine Learning in CodeEasy

How Much Each Direction Holds

A running club measured its members: [height_cm, arm_span_cm, shoe_size_eu, weekly_km]. Height, arm span and shoe size all say roughly how big someone is, so some of these columns overlap.

PCA tells you how much of the variation each new direction carries.

Task: write variance_shares(rows, n_components).

Scale the columns with StandardScaler().fit_transform(...).
Fit PCA(n_components=n_components) on the scaled data.
Return explained_variance_ratio_ as a list of plain floats, each rounded to 3 decimal places.

Examples

rows = [[171.5, 171.8, 40.0, 79.3], [175.3, 174.7, 41.5, 59.0], [175.1, 177.9, 41.5, 57.5], [167.9, 169.3, 39.8, 68.9], [154.2, 155.4, 35.9, 47.9], [176.4, 179.6, 41.9, 55.9], [174.2, 175.5, 41.2, 76.1], [178.1, 180.0, 41.9, 41.0], [175.6, 177.0, 41.2, 49.7], [178.3, 180.2, 41.6, 40.7], [167.6, 170.0, 39.6, 70.0], [175.3, 176.3, 41.4, 59.6]]

n_components = 2

returns [0.75, 0.244]

rows = [[172.9, 174.8, 41.3, 73.3], [164.6, 171.0, 38.7, 41.2], [151.1, 163.8, 36.9, 64.1], [172.2, 170.2, 39.5, 69.0], [164.0, 164.2, 39.4, 47.9], [174.4, 173.2, 41.5, 50.1], [171.9, 180.8, 40.7, 50.8], [180.0, 182.1, 41.6, 54.2], [165.7, 171.0, 40.1, 67.2], [184.2, 186.5, 43.3, 88.8], [173.9, 169.9, 39.1, 58.3], [166.8, 167.4, 40.5, 66.7], [174.5, 172.5, 40.4, 67.9], [163.2, 165.5, 38.1, 65.1], [173.5, 180.3, 42.1, 30.7]]

n_components = 4

returns [0.677, 0.247, 0.043, 0.033]

Hints

solution.py

1from sklearn.decomposition import PCA
2from sklearn.preprocessing import StandardScaler
3
4
5def variance_shares(rows, n_components):
6    pass
7