Looking back at what we have achieved

As you know, up to now, we have benchmarked our code using a subset of the data that contains only the first 100 observations. However, as we saw at the beginning of the chapter, performance can vary for different implementations, depending on the size of the input. To bring together all our efforts in the chapter, we will create a couple of functions that will help us measure how the execution times for our implementations change as we use more observations from our data. 

First, we bring our requirements into R, mainly, the microbenchmark and ggplot2 packages and the files that contain our implementations.

Next, we create the sma_performance() function that takes a symbol, a period, the original_data, a list named sizes whose elements are the number of observations that will be taken from original_data to test our implementations, a cluster to avoid the overhead of initializing it within our sma_parallel() function as we saw in the corresponding section...