Infer information from Tweets. Useful for human-centered computing tasks, such as sentiment analysis, location prediction, authorship profiling and more!
Having a histogram plot of the confidence values predicted for the test set instances could potentially give some insight.
Is the classifier only predicting labels with very high confidence, or is there some degree of uncertainty? What does this say about the classifier (good thing, or not)?
Having a histogram plot of the confidence values predicted for the test set instances could potentially give some insight.
Is the classifier only predicting labels with very high confidence, or is there some degree of uncertainty? What does this say about the classifier (good thing, or not)?
Idea was originally recorded 3/10/13 at 05:36.