10,807 reputation
55095
bio website moderntoolmaking.blogspot.com
location San Francisco, CA
age 28
visits member for 4 years, 3 months
seen 1 min ago

Apr
9
awarded  Popular Question
Apr
5
awarded  Famous Question
Mar
26
awarded  Necromancer
Mar
23
awarded  Constituent
Mar
17
awarded  Caucus
Mar
15
awarded  Notable Question
Mar
14
awarded  Good Answer
Mar
12
awarded  Nice Answer
Mar
11
awarded  Enlightened
Mar
11
awarded  Nice Answer
Mar
10
awarded  Good Answer
Mar
9
awarded  Popular Question
Mar
6
awarded  Notable Question
Mar
5
comment Timeseries analysis procedure and methods using R
@forecaster Also, I'd love for you to check out my cv.ts package on github and leave me feedback on it. I want time series cross-validation to be easy!
Mar
4
comment Timeseries analysis procedure and methods using R
@forecaster Thank you! Notice that ets chose the random walk model as well. At the end of the day I agree: the random walk model is the most appropriate for the data. (the "naive" model is essentially the same as the "random walk" model)
Mar
4
revised Timeseries analysis procedure and methods using R
deleted 32 characters in body
Mar
4
revised Timeseries analysis procedure and methods using R
added theta forecasts, removed some mod_
Mar
4
revised Timeseries analysis procedure and methods using R
added theta forecasts, removed some mod_
Mar
4
comment Timeseries analysis procedure and methods using R
@Niranjan See my new edits. Your dataset doesn't have an obvious trend or seasonal pattern: it's probably a random walk, which makes it very hard to forecast. Also, it's dangerous to choose a model based on a single test set =D.
Mar
4
revised Timeseries analysis procedure and methods using R
added 3656 characters in body