Yesterday, @jsa-aerial and I recorded a short conversation https://youtu.be/DqVh7VEzQaI about Saite, and how it can be used to visualize COVID-19 data.
Hey, total noob here! I'd like to create a simple visualization using clojure of peak loads on a system given a population of users.
I've been messing around with Incanter (just the first google hit i found) and just rendered some normal distributions there.
I'm curious about a few things around this modelling and visualization. 1. Is it a decent assumption to make that usage somewhat fits a normal distribution?
Thanks @val_waeselynck! I watched the video and dived into some telecom whitepapers which uses Poisson distributions for modelling peak call volumes which seems reasonable
Note that the normal distribution is a limit of Poisson distributions, so it may be fine. If you want to do some prediction or interpolation, using normal distributions could give you the opportunity to do that with Gaussian Processes.
Yeah I think it's going to be mostly the same since we just pretty much want to prepare for a bunch of somewhat independent users using a service after it's announced avaliable
Or it's enough of a model to do what we have time to do to mitigate the peak traffic load anyway 😄
Hm I want a nice chart to make a case for progressive rollout of my little service, and visualize it in a histogram with multiple poisson distributions showing cumulative load
I think Zach Tellman made better recommendations here: https://youtu.be/1bNOO3xxMc0
A Poisson distribution iirc