Merge prediction, evaluation and validation code

As a non-researcher, it was difficult for me to distinguish between prediction, evaluation and validation.

1. The code for evaluation and validation is almost identical. Wouldn't it be better to have the same base function? That would:

• avoid potential bugs/errors that would give different results between evaluation and validation,
• make the code easier to read and maintain,
• avoid duplicated code.

2. The code between prediction and evaluation/validation is completely different. Wouldn't it be better to have a common base and use the prediction code in the evaluation/validation code? This would:

• avoid potential bugs/errors that would give different results between prediction and evaluation/validation,
• make the code easier to read and maintain,
• avoid duplicated code that I'm going to do with #252 (closed).