###### tags: `Macine learning`

# Evaluation Matrix[Accurecy and error mesures]

- Evaluation matrix explains performence of matrix
- 3 types of evaluation matrix
1. Jaccard Index
2. F.score
3. Log loss
- Best wat to look the accuracy of classifier is to look at the "confussion matrix"
- confussion matrix shows Actual/true value in row and column shows predicted values
- based on count of each section we can caliculate Precision and Recall

## Actual value


- Precision is measure of accurecy

- $precision = \frac{True \ positive}{Total \ number \ of \ predicted \ +ve}= \frac{True \ positive}{TP+FP}$

- Recall is measure of actual +ve rate

- $Recall = \frac{True \ positive}{True \ positive+False\ Negative} =\frac{TP}{Actual +ve}$

## Accurecy , Precision, Recall(sencitivity), Specificity

$Accuracy = \frac{TP+TN}{TP+FP+FN+FN}= \frac{All\ corctly\ predicted}{All\ sanples}$

$Precision = \frac{TP}{(TP+FP)}= \frac{True \ positive}{True \ positive+False\ Negative}= Tripple \ p \ rule$

$Recall(sencitvity) = \frac{TP}{(TP+FN)}$

$Recall(sencitvity) = \frac{TP}{(TP+FN)}= \frac{TN}{(TN+FP)}$