|
Hi, CompSci grad here, I've started out in performance management (KPIs in particular) and wanted to ask what are the best techniques available to use in the ML world? I've managed to cobble together some raw SQL to build a linear regression line from some sample data so I can plot, but would like to venture towards more predicting the future and other more interesting tools. So, the sample data I've used is; 01/6/11 = 0 08/6/11 = 1 15/6/11 = 2 22/6/11 = 1 29/6/11 = 4 Can ML, given a larger sample, be able to figure out what the future holds? Is there any other techniques that are useful to apply in a performance management system? This is all part of a possible Open Source web based performance app I am thinking of writing. Any help would be greatly appreciated! |
|
ML is designed to solve problems like yours, it is however problem dependent if it can: you might have chosen the wrong model, have access to too little data or just not tried the right set of hyperparameters yet. It's hard for an outsider of performance analysis to guess what problems you are trying to solve. So I suggest you either do some more looking into this yourself or describe what kinds of problems you want to solve in detail. A good introduction is Andrew Ng's lecture at Stanford -- the lectures are also available as videos in the iTunes store, and possibly somewhere else. Thanks for the prompt answer and the link, I will take a look. I guess using this example:
(Jun 12 '11 at 06:53)
Dan
|
