search

26medias / timeseries-analysis

NPM Package - Timeseries analysis, noise removal, stats, ...
238 stars 48 forks source link

coeffs NaN #2

Closed maeishoj closed 8 years ago

maeishoj commented 8 years ago

Hey there.

I am successfully running this to do some forecast on some data samples i periodically store.

However I just tried to do the same code that works on larger dataset (60+ datapoints) on a small dataset (8 datapoints) and the coeffs that I calculate like so:

var coeffs = t.ARMaxEntropy({ data: t.data });

contains [NaN, NaN, NaN, NaN, NaN]

Why is that? is it because I use 5 coeffs and my dataset is too small?

26medias commented 8 years ago

Hi,

Your dataset is too small.

roblav96 commented 8 years ago

I've noticed that anything lower than 10 points will result in this error.

Absolutely astounding work here!

26medias commented 8 years ago

Thanks :)

amprodes commented 4 years ago

HI! in this line of code "t.smoother({period:4}).save('smoothed'" what does mean {period:4} ?

muazhari commented 3 years ago

hi @maeishoj, can I get your code to forecast new n values?