Precision-recall curves in TensorFlow

Introduction

Precision-recall curves help us see how well a model finds positive cases without too many mistakes. They show the balance between catching true positives and avoiding false alarms.

When you want to check how good a model is at finding rare events, like fraud detection.

When false positives are costly, such as in medical diagnosis.

When you want to compare models on how well they identify positive cases.

When accuracy is misleading because classes are imbalanced.

When tuning a model to find the best trade-off between precision and recall.

Syntax

TensorFlow

from sklearn.metrics import precision_recall_curve

precision, recall, thresholds = precision_recall_curve(y_true, y_scores)

y_true is the list of true labels (0 or 1).

y_scores are the predicted probabilities or scores for the positive class.

Examples

Calculate precision and recall for a small example with 4 samples.

TensorFlow

precision, recall, thresholds = precision_recall_curve([0, 1, 1, 0], [0.1, 0.9, 0.8, 0.3])

Plot the precision-recall curve to visualize the trade-off.

TensorFlow

import matplotlib.pyplot as plt
plt.plot(recall, precision)
plt.xlabel('Recall')
plt.ylabel('Precision')
plt.title('Precision-Recall Curve')
plt.show()

Sample Model

This code calculates and prints precision and recall values for different thresholds, then plots the precision-recall curve.

TensorFlow

import tensorflow as tf
from sklearn.metrics import precision_recall_curve
import matplotlib.pyplot as plt

# True labels (0 = negative, 1 = positive)
y_true = [0, 0, 1, 1]

# Model predicted probabilities for positive class
y_scores = [0.1, 0.4, 0.35, 0.8]

# Calculate precision, recall, thresholds
precision, recall, thresholds = precision_recall_curve(y_true, y_scores)

# Print the values
print('Precision:', precision)
print('Recall:', recall)
print('Thresholds:', thresholds)

# Plot the precision-recall curve
plt.plot(recall, precision, marker='.')
plt.xlabel('Recall')
plt.ylabel('Precision')
plt.title('Precision-Recall Curve')
plt.grid(True)
plt.show()

Important Notes

Precision-recall curves are especially useful when the positive class is rare.

The curve shows how precision and recall change as you change the decision threshold.

Higher area under the precision-recall curve means better model performance.

Summary

Precision-recall curves help evaluate models on positive class detection.

They show the trade-off between precision (correct positive predictions) and recall (found positives).

Use them when classes are imbalanced or false positives matter.