Train Grow and Test

[amritbhanu]

• My initial comparisons was with LDA and BoW (just tf). They still hold true.
• But yes zhes other preprocessing technique of using tf + L2 row normalization is for sure performing better.
• this is with Train 60%, grow 20% and test with 20%.

[amritbhanu]

• Results Comparing tf +l2norm, tuned LDA+l2norm, untuned LDA, bow.