Evaluation and assessment of different delivery strategies for Amazon logistics.
This is a simple project that evaluates the performance of two routing strategies across three different scenarios of delivery locations.
Two of the most obvious routing strategies one can take in delivering is a "Greedy" algorithm where the driver goes to the nearest delivery location from their current location, and the "Random" algorithm where the driver would go to a random delivery location out of all of the remaining locations they would have to travel to. To test the performance of these two algorithms, I created three delivery scenarios in which a driver would deliver to. Scenario 1 is an evenly distributed set of points. Scenario 2 is a set of points that are "clustered", or close together. Scenario 3 is two sets of clustered points separated over a distance. All scenarios contain the same number of points, and all scenarios start at the Depot, point (0,0). The simulation can be run to evaluate the total distance spent and the total time spent for each routing strategy.
The app also includes a statistical analysis that computes the average distance/time, minimum distance/time, maximum distance/time, and average standard deviation of distance/time of both the Greedy and Random algorithms, for all three scenarios. In scenarios 1 and 3, we can see that the average distances and times are generally better for the Greedy algorithm compared to the Random algorithm. The Greedy algorithm does generally better than the Random algorithm in scenario 2, however only by a small margin.