First one — simple_solution() doesn't look on the price and consistently chooses for each courier the closest free order, calculates the company profit by sum over all couriers(order_price - courier_salary). Solution works fast.
Second one — courier_solution() works in similar way, but chooses for each courier not the closest order but the most profitable. Courier work until he can benefit the company. Works fast, result: 263240.
Third one — order_solution() for each order chooses the courier that can provide the company with the best profit. These profits make up the sum. By the way, the order of orders(haha) is defined like the earlier the order is available, the earlier we take this order into consideration. Works fast, result: 258676.
For hard_courier_solution() function best_courier() was added, it builds for each courier the chain of actions for the whole day and computes the profit, chooses from them the best one and adds this chain to events list, courier and orders of this chain becomes unavailable. Works for about 40 minutes, result: 266572.
Fifth one — hard_hard_curier_solution() chooses the best courier based not on the whole chain but only the 3 first actions of the day and adds to event only one action(of course, updates the courier position). Works for about a half of the day, result: 270566.
Compile with make
Makefile has parameters:
DF - name of file with data (e.g. - contest)
OF - name of solution file (e.g. - hard_hard_curier)
DATA_DIRECTORY - path to directory with data
OUTPUT_DIRECTIORY - path to answers directory
DATA_PREFICS - prefix of data file
SOLUTION_PREFICS - prefix of answer files
PYTHON - python interpreter
By default we run hard_hard_couriers, on dataset contest.
We can run hard_couriers on dataset simple:
make DF=simple OF=hard_couriers
So the output will be recalculated. To run check.py on the output:
make check DF=simple OF=hard_couriers