statistics - how to interpret IRStatisticsImpl data in mahout -


i want read irstatisticsimpl data have problems:

my result is:

irstatisticsimpl[precision:0.04285714285714287,recall:0.04275534441805227,fallout:0.0018668022652391654,ndcg:0.04447353132522083,reach:0.997624703087886] 

does meant, got 4% of recommendations (precision) , same level of bad recommendation (recall)?

what should numbers @ best - precision @ 1.0 , recall @ 0.0?

well, definition:

precision represents how many results correct in result set. recall represents probability correct element in test set has selected correct , picked in result set.

to perfect precision , recall should both @ 100%. results , criteria these values must evaluated according domain.

for example if have bucket , bad mushrooms should aim @ 100% precision no matter how low recall. because precision critical health, can leave lot of mushrooms. important thing not eat ugly ones. pick 1 mushroom , 100% precision, if there 4 mushrooms in bucket, recall 25%. ideally if precision , recall 100% means in result set mushrooms , also mushrooms in result set , none leaved in test set.

so values might have different meanings.

sadly results seem ugly, because you're having many false positives , false negatives.

take here.


Comments

Popular posts from this blog

c# - Operator '==' incompatible with operand types 'Guid' and 'Guid' using DynamicExpression.ParseLambda<T, bool> -