Neural network for estimating the time spent by programmers. What can be the parameters?

Question

Programming

Hello!
There was a need to soberly assess the time spent on doing work related to programming. For some time now, a table has been maintained by Task Type, Solution Type, Time Estimation, Real Costs. Each major task is divided into as small as possible so that it can be described by a specific Type.
Now, in order to make an assessment based on past experience, you need to extract all the tasks that correspond to the one you need and, as a result, we will get the average of the values. Now the questions are:
1. Can this table be passed as input to a neural network?
2. If you are a programmer or you also need to evaluate a task, what other parameters would you add for an adequate evaluation?
3. Have you seen practical implementations of such functionality?

Answer 1

You can transfer anything, the problem even looks solvable)))
but the easiest way is to sum up (max)s, multiplying the result by the coefficient T,
that is, the usual statistical problem

Answer 2

If the tasks are typical, is it not easier then to teach the neural network to write ready-made code right away?
If the tasks are not typical, then it is unlikely that the neural network will help you in the assessment.

Answer 3

For example , look at https://docs.microsoft.com/ru-ru/dotnet/machine-le... We
download from the description of taxi-fare-trans.csv, count the number of lines... compare with the number of half of our known facts ( in order to feed the soulmate for the test) ... we think ...