Update a Sentiment Analysis Model

ang30ang30 Member Posts: 1 Contributor I
edited September 2019 in Help

Hi,

 

I built a sentiment analysis model by using 1000 sampling data.

Now, I have additional 1000 examples, and I would like to update the model.

 

I know that I can input those 2000 example records and train the model again.

However, it is not efficient to do so.

 

Is there any method to update the trained model by new sampling?

Thanks!

 

Answers

  • lionelderkrikorlionelderkrikor Moderator, RapidMiner Certified Analyst, Member Posts: 1,193 Unicorn

    Hi @ang30,

     

    I propose you the following solution : 

    Is this new process more efficient ? 

    Regards,

     

    Lionel

     

    Update_model.png

     

    Thomas_Ott
  • sgenzersgenzer 12Administrator, Moderator, Employee, RapidMiner Certified Analyst, Community Manager, Member, University Professor, PM Moderator Posts: 2,959 Community Manager

    huh - been using RapidMiner for five years and have never seen "Update Model". Thank you @lionelderkrikor. Learn something every day!


    Scott

     

    Thomas_Ottlionelderkrikor
  • lionelderkrikorlionelderkrikor Moderator, RapidMiner Certified Analyst, Member Posts: 1,193 Unicorn

    Hi,

     

    You're welcome @sgenzer. Chance would have it : I discovered this operator yesterday .......

     

    Regards,

     

    Lionel

    sgenzer
  • Thomas_OttThomas_Ott RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,760 Unicorn

    Just one caveat with it, it will only work if the model generate can be updatable. The Deep Learning algo is not updatable whereas Naive Bayes is.

    sgenzer
  • Telcontar120Telcontar120 Moderator, RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,630 Unicorn

    Exactly, it's a nice option in theory, but unfortunately most popular algorithms don't work with it: decision trees (and their variants), linear and logistic regression, neural nets, SVMs, k-NN, etc., are all not updatable.  In fact, Naive Bayes is about the only common one I can think of that is!

    Brian T.
    Lindon Ventures 
    Data Science Consulting from Certified RapidMiner Experts
    sgenzer
  • charlycharly Member Posts: 6 Contributor I
    helo dears ,, whats the new about this  ( update model ) in rapid miner ??
     
Sign In or Register to comment.