Kaggle competition: Partly cloudy with a chance of hashtag

This is my code for a recent kaggle's competition hosted here.
http://www.kaggle.com/c/crowdflower-weather-twitter

required python modules:
=======================
gensim
nltk
scikit-learn
numpy
scipy
pandas

How to run the code?
- Download train.csv and test.csv on the competition's website
- run either train or train_model in train.py