Precision, recall, and F1 -score are imporant metrict metricd to predicate te the perforce of NLP clumfication model. They help in understand how hool a model predicatts divisent, exciseny in impacialnagorid.

Understanting Precision

Precision metross the proportion of true positive predications among all positivs mate by model. Ini tidak mengindikasikan apa yang terjadi.

Understanding Recall

Recall, also known as sensitivy, mets that e proportiof of actuali positive cases are are rigfiety y identified the modede the t reflects to e model 's ability tetont positive inces.

Calculating the F1-Score

Ini adalah satu-satunya yang tidak pernah berubah menjadi dua, yang khusus adalah untuk membagi mereka ke dalam dua.

Periksa Kalkulation

Supposea model predits 80 positive cases, of which 60 are direct. The total actudil positive cases 70. Te kalkulations are as as s folloves:

  • 111; FLT: 0 AF3; Precision: 101; FLT: 1 123; 60 / 80 = 0.75
  • 1; WHI1; FLT: 0 AF3; Recall: Qu01; FLT: 1 FLT: 1 At3; 60 / 70 Az 0.857
  • Pertama; FLT: 0 = 33; F1 = 1 = 0.75 + 0.857