Figure 5: Evolution of execution time with a varying percentage of missing values in the small (a) and large (b) datasets.