This is the simulation code for our work "Online Algorithms for Multi-shop Ski Rental with Machine Learned Advice" accepted by NeurIPS 2020.
All codes are in Matlab format. You can simply run the corresponding code to get the desired figure. A detailed decription of the figures will be given below.
Figure 1 shows the performance of single prediction, multi-shop deterministic algorithm( Algorithm 2).
Figure 2 shows the performance of single prediction, multi-ship randomized algorithm( Algorithm 3).
Figure 3 shows the possible outperformance of Algorithm 2 over Algorithm 3.
Figure 4 shows the impact of \lambda.
Figure 5 shows the impact of biased errors for both Algorithm 2 and 3.
Figure 6 gives a real-world dateset to simulate the performance of Algorithm 2, compared with pure online algorithm without predictions.
Figure 7 shows the performance of multi-prediction, multi-shop deterministic algorithm( Algorithm 4).
Figure 1 shows a different setting of \gamma which gives a different performance in Algorithm 2 and 3.
Figure 2,3,4,5 show a wider range of \lambda and \gamma for biased error for Algorithm 2 and 3.
Figure 6 shows different performance for different \lambda using Algorithm 4.
Figure 7 shows different performance for different number of predictions using Algorithm 4.
Figure 8,9 show the impact of biased error using Algorithm 4.
Figure 10,11 show the performance of multi-prediction, multi-shop randomized algorithm( Algorithm 5).
Figure 12 shows the performance of Algorithm 4 with real-world dataset.
Any academic work, which is built on this code, should use reference of the following paper.
Online Algorithms for Multi-shop Ski Rental with Machine Learned Predictions
Shufan Wang, Jian Li, Shiqiang Wang
NeurIPS 2020