New answers tagged

0

I looked for this as well and found only the unmaintained RAT library. But basically all of these algorithms (most are described as formulas in this paper) can be directly translated from the single Python file with no extra dependencies. One exception was the Linsear Write formula because of its needing a syllable counting method. But if you use the CMU ...


0

For the tags: Do you know how they are generated? How many unique tags do you have? If they are self-generated (ie lots of tags that can be subsets of other tags). You might need to do tag consolidation which will also help with dimensionality reduction of the word vectors. If you can provide a bit more information about what your data looks like and where ...


0

It is generally best practice to perform all feature engineering before applying classifiers. The two primary reasons are: Simplicity - If feature engineering is conditional on model performance then it is harder to find and debug edge cases. Handle of sample issues - Especially in text, there are novel examples (e.g., words that appear during prediction ...


Top 50 recent answers are included