📏 Micro Precision

Micro Precision#

Precision computed from globally pooled true positives and false positives.

Important

✨ AI-generated content. This page was written with the assistance of an AI language model and is provided as a learning aid. Despite careful review, it may still contain mistakes, omissions, or out-of-date information. Whether you are new to the topic, a team lead, or a senior practitioner, treat it as a starting point rather than an authoritative reference: read it critically and independently verify anything you act on (code, commands, figures, and factual claims) against official documentation and primary sources before relying on it.

What it is#

Precision answers: of all the items the model flagged positive, how many were right?

\[\text{Precision} = \frac{TP}{TP + FP},\]

so it is sensitive to false positives (false alarms). Micro precision extends it to \(K\) classes by pooling counts before dividing, rather than averaging per-class precision (macro precision).

How it’s computed#

\[\text{Precision}_{\text{micro}} = \frac{\sum_{i=1}^{K} TP_i}{\sum_{i=1}^{K} (TP_i + FP_i)}.\]

Summing across classes first makes the metric a single global “of all predictions, how many correct”, so frequent classes carry the most weight.

The micro identity#

As with micro recall and micro F1, in single-label classification micro precision equals the other two (and accuracy), because the pooled denominators line up. They differ only under multi-label evaluation.

Worked example#

Three classes with TP = (40, 30, 10), FP = (10, 20, 20):

Precision(A)=0.80, Precision(B)=0.60, Precision(C)=0.33 → Macro precision = 0.58.
Micro precision \(= 80/130 \approx 0.615\).

The micro value sits near the large classes’ contribution; macro surfaces class C’s weaker 0.33.

When precision matters most#

Favour precision when a false positive is costly — spam filters (blocking real mail), recommending a bad product, flagging an innocent transaction — where the price of a wrong “yes” is high.

In code#

from sklearn.metrics import precision_score

micro = precision_score(y_true, y_pred, average="micro")
macro = precision_score(y_true, y_pred, average="macro")

Theme: Classification & Averaging Metrics · All terminology

Hint

Mind map — connected ideas

Micro Recall · Micro F1 · Macro Precision · Micro AUROC

Hint

More in Classification & Averaging Metrics

Accuracy · AUC (Area Under the Curve) · Average Precision (AP) · Binary Classification · Classification Probability · Discriminatory Power · F1-score · Gini Coefficient · Harmonic Mean · Log Loss (also called Logarithmic Loss or Cross-Entropy Loss) · Macro AUC · Macro AUROC (Macro-Averaged AUROC) · Macro Averaging · Macro F1