Precision and recall are often used in machine learning, recommendation systems, information retrieval, natural language processing, multimedia vision, and other fields) f-measure, F1-score to evaluate the accuracy of the algorithm.
I. Accuracy and recall rate (P & R)
Take text search as an example.
The black box indicates the search domain, from which we retrieve items that are highly correlated with the target text. The yellow part (A + B) in the figure indicates the items with a high priority between the domain and the target text.
A + C indicates the items retrieved by your algorithm. The meanings of A, B, and C are marked in Chinese and English.
Accuracy:
Recall rate:
In general, accuracy indicates how many items are correctly retrieved by your algorithm, and recall rate indicates how many items are retrieved accurately.
Relationship between accuracy and recall rate
Generally, we hope that the higher the accuracy and recall rate, the better, but in fact the two are in conflict in some cases. For example, if only one result is found, the result is correct and the precisin value is 1. However, because only one result is found, the recall value is very low, close to 0. Therefore, the following section describes F-measure.
II. F-Measure
F-measure is also called F-score. Its formula is:
When Beta = 1, it is F1-score:
F-measure combines precision and recall. The higher the value, the better the algorithm performance.
Precision & recall)