Downsample

The downsampling tool takes a parent population and creates a new population that has a subset of events from the parent population. This can be useful if you want to run an algorithm such as tSNE, which take a longer time if the number of events in the population is higher. Downsampling the population before running tSNE can significantly decrease the amount of time the algorithm takes to run.

Create Downsample

To downsample a population, select the population and then click the "Sample" tool from the toolbar. This will bring up a pop up where you can select the amount of events that the new population will have. You must sample an event count that is greater than 0 and less than or equal to the number of events in the parent population. Then you can select "Create" and a new child population will be created with that number of events.

How Downsampling is Done

There are a few different ways that downsampling can theoretically be done. For example, you can randomly select events, sample at a constant interval, or sample based on event density.

Currently in Floreada.io, events are sampled from the parent population at a constant interval which is determined by the ratio between the parent and downsampled event counts. For example if the parent population is 100000 and you have selected a downsampling event count of 10000, the downsampling process will take every 10th event in the population to include in the downsampled population.