Which data fitting model to use for this problem

Question

My data set has 15 independent variables and 2 predictors or response. The problem is to find a mapping between input and output variables. Basically, given a feature vector as input, the trained model will give the estimated or the predicted value of the response. My input data consists of chemical concentration, yield of the crop, percentage rainfall etc and using this input the model should be able to predict the phvalue of the soil and temperature. The data is collected over a year.

During training the model should be able to exactly replicate the response. Then the trained model should be able to output the response of a day given an input feature for that day.

I have split the data -- 70% I have used for training and 30% for testing. My idea was to use LSTM Matlab document. Can LSTM be used or is there a better model for this task?

My confusion is that in all documents LSTM is applied to predict sometime ahead in future. But my problem leans toward fitting the data to a function. I have tried using regression but the accuracy is very poor. Please suggest what are other methods suitable for this problem.