-2

I am working on testing accuracy and performance using deep learning models on a complex dataset but I have reached a good accuracy but I need to improve it so any suggestions other than what I did(feature selection ,Information Gain, Recursive feature elimination (RFE) , Random Forest Important Scoring, I have also used SMOTENN and it was the best for this imbalanced dataset so any other approach anyone could suggest

I did(feature selection ,Information Gain, Recursive feature elimination (RFE) , Random Forest Important Scoring, I have also used SMOTENN

Increasing Accuracy

Improve this question
1
  • 1
    This is not a programming question, it is off-topic in Stack Overflow. Commented Dec 20, 2024 at 22:03

1 Answer 1

Reset to default
2

To help you more we would need to understand the following things:

  1. What type of data are you using? How much data is there? What's the quality of the data?

  2. What are you trying to predict?

  3. Is accuracy the right measure for the task?

  4. What accuracy have you reached so far (getting from 40->50% accuracy is easier than 90->95% accuracy)

In general I'd say this is a difficutl question to answer without a lot more information.

General tips: Try a different deep learning model Try a different loss function Check for overfitting Try training for more epochs Try a different optimiser Look at individual failure cases, work out what the model is doing wrong and design your own tweaks to handle those cases.

Improve this answer
Sign up to request clarification or add additional context in comments.

2 Comments

Hi @MattLovesJam It is NB15 dataset highly imbalanced dataset I have reached 99 but need to improve it or get another approach that could get more closer accuracy to 99 , I tried MLP CNN, I have even thought to re-imbalance that dataset after balancing it but that approach didn't get me a good accuracy, so any other approach do you suggest
The question is off-topic due to not being about programming, you are not supposed to try to answer off-topic questions!

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.