Thakshashila: /* SEO Keywords */

2025-06-10T06:23:08Z

SEO Keywords

← Older revision		Revision as of 06:23, 10 June 2025
Line 85:		Line 85:

	micro f1 score, f1 score for multi-class classification, machine learning evaluation metrics, micro average f1, macro vs micro f1, multi-label classification f1 score, f1 score imbalance		micro f1 score, f1 score for multi-class classification, machine learning evaluation metrics, micro average f1, macro vs micro f1, multi-label classification f1 score, f1 score imbalance
			[[Category:Artificial Intelligence]]

Thakshashila: /* Related Pages */

2025-06-10T05:49:27Z

← Older revision		Revision as of 05:49, 10 June 2025
Line 75:		Line 75:

	* [[F1 Score]]		* [[F1 Score]]
	* [[Macro F1 ~~Score~~]]		* [[Macro F1 ]]
	* [[Weighted F1 ~~Score~~]]		* [[Weighted F1]]
	* [[Precision]]		* [[Precision]]
	* [[Recall]]		* [[Recall]]

Thakshashila: Created page with "= Micro F1 Score = The '''Micro F1 Score''' is an evaluation metric used primarily in '''multi-class''' and '''multi-label classification''' tasks. Unlike Macro F1 Score, it calculates global counts of true positives, false positives, and false negatives across all classes, then uses these to compute a single Precision, Recall, and F1 Score. It is most useful when the dataset is '''imbalanced''' and you care more about overall performance than per-class fai..."

2025-06-10T05:26:33Z

Created page with "= Micro F1 Score = The '''Micro F1 Score''' is an evaluation metric used primarily in '''multi-class''' and '''multi-label classification''' tasks. Unlike Macro F1 Score, it calculates global counts of true positives, false positives, and false negatives across all classes, then uses these to compute a single Precision, Recall, and F1 Score. It is most useful when the dataset is '''imbalanced''' and you care more about overall performance than per-class fai..."

New page

= Micro F1 Score =

The '''Micro F1 Score''' is an evaluation metric used primarily in '''multi-class''' and '''multi-label classification''' tasks. Unlike [[Macro F1 Score]], it calculates global counts of true positives, false positives, and false negatives across all classes, then uses these to compute a single [[Precision]], [[Recall]], and F1 Score.

It is most useful when the dataset is '''imbalanced''' and you care more about overall performance than per-class fairness.

== Definition ==

:<math> \text{Micro F1} = 2 \cdot \frac{\text{Micro Precision} \cdot \text{Micro Recall}}{\text{Micro Precision} + \text{Micro Recall}} </math>

Where:

:<math> \text{Micro Precision} = \frac{\sum TP}{\sum TP + \sum FP} </math>
:<math> \text{Micro Recall} = \frac{\sum TP}{\sum TP + \sum FN} </math>

Instead of averaging individual class scores, Micro F1 aggregates global totals of:

* True Positives (TP)
* False Positives (FP)
* False Negatives (FN)

== Step-by-Step Example ==

Suppose a 3-class classification problem with:

* Class A: TP=50, FP=10, FN=5
* Class B: TP=30, FP=15, FN=10
* Class C: TP=20, FP=5, FN=15

Global totals:
* TP = 50 + 30 + 20 = 100
* FP = 10 + 15 + 5 = 30
* FN = 5 + 10 + 15 = 30

:<math> \text{Micro Precision} = \frac{100}{100 + 30} = \frac{100}{130} \approx 0.769 </math>
:<math> \text{Micro Recall} = \frac{100}{100 + 30} = \frac{100}{130} \approx 0.769 </math>

:<math> \text{Micro F1} = 2 \cdot \frac{0.769 \cdot 0.769}{0.769 + 0.769} = 0.769 </math>

Micro Precision and Recall are equal, so Micro F1 equals them.

== Micro vs Macro vs Weighted F1 ==

{| class="wikitable"
! Metric
! How It Works
! Best For
|-
| '''Micro F1'''
| Global average across all classes (TP, FP, FN summed first)
| Imbalanced data where you care about overall performance
|-
| '''Macro F1'''
| Average of F1 scores per class (unweighted)
| Equal importance for each class
|-
| '''Weighted F1'''
| Average of F1 scores per class (weighted by class size)
| Imbalanced data, focus on majority classes
|}

== Use Cases ==

* Multi-label text classification
* Image tagging tasks
* Medical diagnosis systems with multiple labels
* Imbalanced datasets with focus on global accuracy

== Limitations ==

* May **hide poor performance** on minority classes
* Doesn't reflect per-class fairness

== Related Pages ==

* [[F1 Score]]
* [[Macro F1 Score]]
* [[Weighted F1 Score]]
* [[Precision]]
* [[Recall]]
* [[Accuracy]]
* [[Confusion Matrix]]

== SEO Keywords ==

micro f1 score, f1 score for multi-class classification, machine learning evaluation metrics, micro average f1, macro vs micro f1, multi-label classification f1 score, f1 score imbalance

Micro F1 Score - Revision history

Thakshashila: /* SEO Keywords */

Thakshashila: /* Related Pages */