Undergrad minor (7th sem) project on classification of webpages as Evergreen or Ephemeral based on their webpage content
Contest Link : http://www.kaggle.com/c/stumbleupon
Techniques Used:
-
RandomForest using all fields except body
- 20 Fold CV Score : 80.7915%
-
- Logistic Regression on Tf-Idf vectorized body
- 20 Fold CV Score : 87.7833%
- Logistic Regression on Tf-Idf vectorized body with Kstratfold and SelectPercentile using chi after outlier removal
- 20 Fold CV Score : 89.15924%
- Logistic Regression on Tf-Idf vectorized body
-
Gaussian Naive Bayes using all fields except body
- 20 Fold CV Score : 70.379%
-
Linear SVM on Tf-Idf vectorized body
- 20 Fold CV Score : 86.8915%
Tf-Idf was done along with stemming and tokenizing to improve accuracy
- PunktWordTokenizer
- SnowBallStemmer
- LemmaTokenizer (see LRwithchitest.py)
To Use please place train.tsv and test.tsv in the same directory and run any of the files