Accuracy (Precision), recall rate (Recall) and comprehensive evaluation Indicators (F1-MEASURE)

Accuracy and recall are the two concepts and indicators often involved in data mining prediction, Internet search engine and so on.

Accuracy Rate : Also known as "precision", "correct rate"

Recall rate : Also known as "recall"

Take the search as an example, can be used to express the situation:

	Related	Not relevant
Retrieve the	A	B
Not retrieved	C	D

A: retrieved, related

B: retrieved, but not relevant

C: Not retrieved, but is relevant

D: Not retrieved, also irrelevant

1. Recall rate: The more content retrieved, the better, i.e. r=a/(A+C)

2. Accuracy: In the retrieved document, the more you really want, the better, i.e. p=a/(a+b)

In different situations, it is necessary to judge whether the P is higher or r is higher. If you are doing experimental research, you can draw Precision-recall curve to assist analysis.

f1-measure:P and R indicators are sometimes contradictory and require some way to consider them comprehensively.

F-measure is a weighted harmonic average of P and R: f= ((a2+1) p*r)/(A2 (P+r))

When the parameter a=1, it is the most common F1: f1=2pr/(p+r)

F1 synthesized the results of P and R, when the F1 was higher, the experimental method was more ideal.

Accuracy (Precision), recall rate (Recall) and comprehensive evaluation Indicators (F1-MEASURE)