Hello all,
I want to go for publication in a journal related to medical prediction, but my proposed machine learning algorithm doesn't show great results.
Results of the proposed method are better than existing methods( when tried with different existing classifiers) but they are not very good when seen alone.
My proposed algorithm looks good, with a very novel approach. I just want to ask if results could be an issue while accepting my paper, and how should I proceed.
thanking in anticipation.
If you have developed a novel approach which gives better performance than the existing works for your problem statement (or dataset) then you can publish your work.May be you can come up with some kind of optimization to further improve your results
I would say that the approach needs to be determined to be robust, before you can publish it - you need to be sure that it works the way it should, based on several realistic test cases. It also has to be compared with state-of-the-art methodology, again using several realistic test cases, and the verdict then needs to be clearly in favour for your new approach - that you get better predictions, and/or faster - if that is an issue.
Himani Deshpande i think that you should go for machine learning focused journal first of all. And, secondly you could include optimization like feature extraction and feature selection. Lastly, third you can include more dataset and compare the performance of the algorithm like this one:
Preprint Implementing Naive Bayes Algorithm for Detecting Spam Emails...
I would agree to broaden your dataset and in parallel perform tests with several p levels. You could see your algorithm working better on a different level and maybe that is not so weak. the number of data should be chosen in relation to the confidence interval. If you don't manage to attain strong results do not publish your work.
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication