100 accuracy recall and precious

fatimidveil

i have apply ide algorithm on my data set and got 100 accuracy precious and recall . is my data set and algorithm in fine ?

varunm1

Hello @fatimidveil

100% accuracy is something one should be very careful about. You need to check a few things listed below.

1. Label leakage: If your data set is time-dependent (temporal characters), using regular validation methods like cross-validation is not recommended. 
2. Highly correlated column: If your dataset has an attribute that replicates output, then there is a chance of getting 100% accuracy.
3. Validation Type: If you are using something like a split validation, there is a chance that you might get very high random performance.
4. Size of a dataset: If your dataset is too small and you have a very small test set, then you can get very high performance by chance.

These are some points in my mind right now, will update if I get more.

fatimidveil

hi everyone,as you mention above. My data set consist of 1150 entities and i have one attribute that is highly correlated with my class attritube ..
what should i do know .?
i have apply three algorithm on my data set id3,cart and c4.5 
so how i calculate with one is perform better than other ?