Where Do You Draw the Line? — easy Machine Learning in Code problem | Incognition

Where Do You Draw the Line?

10 pts · 10 coins

Machine Learning in CodeEasy

Where Do You Draw the Line?

A fraud model gives every payment a chance of being fraud. The bank flags a payment when its chance is at or above a threshold it chooses.

Task: write flag_and_score(y_true, chance, threshold) and return [precision, recall], each rounded to 3 decimal places.

y_true holds 1 for fraud and 0 for a normal payment. It contains at least one 1.
chance holds the model's chance of fraud for each payment, in the same order.
Precision: of the flagged payments, the share that really were fraud.
Recall: of the real frauds, the share that were flagged.
If nothing is flagged, precision is 0.0.

Try a high and a low threshold on the same payments and watch the two scores pull against each other.

Examples

y_true = [1, 0, 1, 0, 0, 1, 0, 0]

chance = [0.9, 0.6, 0.4, 0.2, 0.1, 0.7, 0.3, 0.05]

threshold = 0.5

returns [0.667, 0.667]

y_true = [1, 0, 1, 0, 0, 1, 0, 0]

chance = [0.9, 0.6, 0.4, 0.2, 0.1, 0.7, 0.3, 0.05]

threshold = 0.25

returns [0.6, 1.0]

Hints

solution.py

1import numpy as np
2from sklearn.metrics import precision_score, recall_score
3
4
5def flag_and_score(y_true, chance, threshold):
6    pass
7