I am currently trying to find a solution to finding outputs to analytical equations, for example:
Y= 1/(X1*X2); where X1 and X2 are input dataset and Y is the output that is expected from the algorithm.
The main point is not using the analytical expression to solve equations but rather passing X1 and X2 as inputs in the form of a dataset and using the algorithm to predict Y. I have tried using neural network and a few other algorithms on this dataset, however, I am getting very high MSE.
I need a suggestion as to what algorithm to use or what approach to use.
Dear Mahnoor Khan ,
earning data science may seem intimidating but it doesn’t have to be that way. Let’s make learning data science fun and easy.
https://towardsdatascience.com/how-to-build-a-machine-learning-model-439ab8fb3fb1
Regards,
Shafagat
You may increase the training instances and also try to tune the hyperparameters used for the selected algorithm.
Choose the suitable features of X1 and X2 for training. Check its accuracy. If accuracy is good, then prediction will be good.
You may have to remove redundancy and clear data of inconsistencies. During training adaptively optimise the parameters to reduce error.
For each type of equation, it can be assumed that the solutions depend on the values of the coefficients. It is necessary to give examples of how exactly they depend. after that, there is everything to train the system to predict decisions based on coefficient values. If you send training and test samples, then I can use the intelligent Eidos system that I developed to solve your problem and write what the reliability of the models according to van Rizbergen's F-measure turned out to be.
you can used traditional machine learning or deep learning methods through nonlinear function to model the given equation, it's okay
you can use another metric such as MAE, RMSE, also extend dataset, use more complex model, normalize or standardize dataset
- Badges
- Science method