Easy

Have a go at it. The editor and the docs panel are open, and your code is saved as you type. Running it needs a free account — you’ll come back to exactly what you wrote.

Classification Metrics

~12 mincode completion

Accuracy is the fraction of predictions that were right. It is also, on imbalanced data, close to useless: if 1 percent of transactions are fraud, predicting "not fraud" every single time scores 99 percent accuracy and catches nothing.

The two that actually describe a classifier:

Precision: of the cases you flagged, what fraction really were positive? Low precision means you are crying wolf.

Recall: of the cases that really were positive, what fraction did you catch? Low recall means you are missing them.

precision = \frac{T P}{T P + F P} recall = \frac{T P}{T P + F N}

They trade off. Flag everything and recall is 1.0 while precision collapses. Flag nothing and precision is undefined while recall is 0. Which one you care about is a question about consequences, not about maths: a missed tumour and a false alarm are not equally bad.

from sklearn.metrics import accuracy_score, precision_score, recall_score

accuracy_score(y_true, y_pred)
precision_score(y_true, y_pred, zero_division=0)
recall_score(y_true, y_pred, zero_division=0)

zero_division=0 says what to do when the denominator is zero, which happens whenever the model predicted no positives at all. Without it you get a warning and a nan.

Your task:

Write score_model(y_true, y_pred) that returns [accuracy, precision, recall], in that order, for binary labels of 0 and 1.

Example Tests

A perfect classifier scores 1.0 on all three

Input: {"y_pred":[0,1,0,1],"y_true":[0,1,0,1]}

Expected: [1,1,1]

One false positive lowers precision but not recall

Input: {"y_pred":[1,1,0,1],"y_true":[0,1,0,1]}

Expected: [0.75,0.66667,1]

One missed positive lowers recall but not precision

Input: {"y_pred":[0,1,0,1],"y_true":[0,1,1,1]}

Expected: [0.75,1,0.66667]

Python

import numpy as np
from sklearn.metrics import accuracy_score, precision_score, recall_score

def score_model(y_true, y_pred):
    """
    Report the three headline classification metrics.

    Args:
        y_true: 1-D array of true labels (0 or 1)
        y_pred: 1-D array of predicted labels (0 or 1)

    Returns:
        A list [accuracy, precision, recall].
    """
    # YOUR CODE HERE
    pass

Run your code to see results

⌘↵ runs against the visible tests

Loading docs…