Something is obviously wrong. Most people would use binary logistic regression for this research. If you are not familiar with that approach google it and much information will result. I have hopefully attached useful information for you. Questions are are always accepted. Best wishes, David Booth

Oops had connection problem here's the attached information. Remember questions are always accepted here. Best wishes and good luck, David Booth

Oops #2 what are the percentages of 0 and I. If vvv different in your data that can cause a problem as well. Also what are your predictors. If we can make this work it should help on your NN model.. Best wishes, David Booth

Hi

DEAR David Eugene Booth

The problem of protein-peptide interaction can be viewed as a binary classification problem, where in a protein chain, each residue can be classified into one of two classes: binding or non-binding . Some techniques that have been used with this strategy include SVM, random forest and artificial neural networks (ANN).

So, I used the RNN neural network as a binary classifier that only predicts zero labels,Unfortunately.

I'm confused because this prediction is wrong

Thank you very much for sending the resources and I will definitely use them to solve my problem

Thank you very much for your guidance

Be healthy

Hello Dear Shima

I have question. How does it predict zeros.

In the logistic regression if we predict the probability of zero, then the prediction of the probability of 1 is 1-P(y=0), if y is the data label.

If it is using the logistic regression model, it will be the case otherwise I would not recognize.

Thank you very much

HELLO DEAR PARISA

I have used deep learning (with Python programming)for predicting. So, the output of the proposed method is 0 value.

for example, an LSTM is a type of deep learning-based method. It has a similar control flow as a recurrent neural network. It processes data passing on information as it propagates forward. The differences are the operations within the LSTM's cells. These operations are used to allow the LSTM to keep or forget information.

LSTM Network for Regression. LSTMs are sensitive to the scale of the input data, specifically when the sigmoid (default) or the activation functions are used. It can be a good practice to rescale the data to the range of 0-to-1, also called normalizing.

I recommend the article on the relationship between regression and deep learning

"Solving Classification and Regression Problems Using Machine and Deep Learning"

• DOI:
• 10.13140/RG.2.2.21723.21288

If you have any questions, ask again , please

GOOD LUCK

Dear Shima

Hello

Thank you for your response. Are you predicting data that are continuous and are between 0 and 1?

Or are they taking only 0 and 1?

I am sorry.

Thanks

Ahh and according to your first question, the data are binary.

So are they using binary regression?

Thank you very much

Because as they have used binary model, as long as we predict the probability of 0, say, the probability of 1 is predicted by 1-p(y=0).

And I have a question. are they predicting the probability of zero, or the number of zeros?

Then it needs to know the function of the model. I suppose.

I am sorry.

Thank you

best