2
I don’t know much about this world of artificial intelligence, but I have a data set that tends to be a linear function, and one thing I needed was to find the angular coefficient of the line that fits this data.
Using linear regression, I was able to find it through MATLAB, but I wanted to use the functions of this software’s neural networks to make a prediction of the values, but what it gave me was a "cloud of points" (I used the nnstart function in Matlab to perform the training)
as follows in the image below, the first figure is linear regression and the second is what I found in the neural network
The only way to find the coefficient that the neural network tried to predict is by what I did or is there some other more efficient method?
See if your neural network is using activation function, if yes, you need to remove it. To use a perceptron for regression you need a linear output.
– Felipe Franco