Generally, an index system can be established to evaluate the problem. The index system can be a set-valued function or a single-valued function. A single-valued function means: z=f(x 1, x2, ..., xn), where xi is the index of each attribute. The system can be evaluated by the value of z, or it can be made into a multi-valued function (z 1, z2, ..., Zn) = f (x 1, x2, ...). Single-valued functions are generally suitable for evaluating a system, while set-valued functions are generally suitable for evaluating and comparing systems.

For the prediction model, a regression function can be established to predict. Of course, the type of regression function should be carefully selected.

In addition, the grey prediction method can be used to predict the system with little regularity.

For some systems with special laws, smoothing method or time series method can be used.

Specifically, I suggest you look at the corresponding reference books!