Method C.

Method C involves analyzing logs(by Default last 10) and extracting from logs completed operations. Extracted operations are being ordered by the time in which they were started, starting from oldest to newest. Each operation is marked as operationalAfter if after ending that operation has ended other operation. In this window between ending operation and ending other operation, all operation that has started are being checked if they succeeded or failed, if there is more successful operations than the original operation is marked as successfullAfter. In this way are obtained vectors of completed operations in row, with values if they were successful or not and value representing if last operation in that row is successfullAfter. For that vectors are being calculated vectors Values that are defined as a sum of all values of operations in vectors multiplaied by modificators specified by some linear function, that is described by value in first operation in vector and value in last operation in vector. There are checked many combinations of this values and it is chosen that combination of values that most correctly divides vectors between those that are successfullAfter and that that are unsuccessfullAfter divided by some border value. That border value and combination of values multiplying first operation and last operation in vector will be used later when ends of live operations that will be determined during Robots work. Value of that kind live vector will be used to specify if operations starting after ending vector will be executed directly or with use of Reverse Operations logic. This operations are being performed with specified amount of values in the row of completed operations(by Default from 3 to 10). For amount of operations that best divide between successfullAfter and unsuccesfullAfter, the Neural Network is being trained with extracted rows of operations.