The First Question — medium Machine Learning in Code problem | Incognition

The First Question

10 pts · 10 coins

Machine Learning in CodeMedium

The First Question

A tree's first question is the single split that separates the classes most cleanly. You want to know which column it asks about.

Write first_question(rows, labels, names).

names holds the column names, in the same order as each row.
Fit DecisionTreeClassifier(max_depth=1, random_state=0).
Use export_text(model, feature_names=names) to get the tree as text. Its first line has the form |--- <column name> <= <threshold>.

Return a dict:

"column": the name of the column the first question uses, as a string.
"threshold": the number in that question, as a float, exactly as export_text shows it.

Examples

rows = [[12, 3.1, 0.4], [30, 2.2, 0.9], [8, 9.5, 1.6], [25, 7.8, 0.3], [5, 1.2, 0.2], [40, 4.4, 1.1], [15, 8.9, 0.8], [22, 1.5, 1.9], [3, 6.1, 0.5], [35, 9.9, 0.6]]

labels = [0, 1, 0, 1, 0, 1, 0, 1, 0, 1]

names = ['sugar', 'fat', 'salt']

returns {'column': 'sugar', 'threshold': 18.5}

rows = [[2.1, 40], [3.5, 22], [1.2, 35], [4.8, 28], [2.9, 18], [5.5, 45], [1.8, 12], [4.1, 50]]

labels = ['no', 'yes', 'no', 'yes', 'no', 'yes', 'no', 'yes']

names = ['rainfall', 'temperature']

returns {'column': 'rainfall', 'threshold': 3.2}

Hints

solution.py

1from sklearn.tree import DecisionTreeClassifier, export_text
2
3
4def first_question(rows, labels, names):
5    pass
6