Skip to content

aakash2602/IdentifyOnlinePatientConversations

Repository files navigation

IdentifyOnlinePatientConversations

Problem Statement: Conversation log of people asking question related to some disease We need to classify the documents into 2 types - conversation related to cancer and all the other conversations.

2 Approachs:

  1. BOW -> LSI -> Logistic Regression
  2. BOW -> TFIDF -> Shallow NN

Results: Approach 2 works best with 90%+ accuracy on Test Data ROC curve is used for evaluation