issues
search
Svetuf
/
CSC_Sentiment_Russian_NER
0
stars
0
forks
source link
Compare quality of several models
#3
Open
roddar92
opened
3 years ago
roddar92
commented
3 years ago
Compare means of our metrics (precision, recall, f-score):
Launch our algorithms (5-10 times) on the test sample
Calculate all previous metrics and stores their results to make statistical analysis:
confidence interval (95%-97%)
hypothesis analysis (null hypothesis - quality of models are the same, alternative - quality is changed), T-test
Compare means of our metrics (precision, recall, f-score):