The business user creates a task and picks out two datasets from uploaded data: one for the scientists, one for testing quality.
Scientists download the dataset and make a algorithm to solve the problem.
DMH evaluates and compares the algorithm quality on the covered dataset and cost of usage setting up by the scientist.
The customer chooses the algorithm and begins the calculation, using all of the Data. When the calculation is finished, the customer receives predictions and the scientist receives the money.
After that the customer starts a new iteration with the original data or uploads new data.