search

alkaline-ml / pmdarima

A statistical library designed to fill the void in Python's time series analysis capabilities, including the equivalent of R's auto.arima function.
https://www.alkaline-ml.com/pmdarima
MIT License
1.58k stars 234 forks source link

About arima parameter m #19

Closed peilj closed 6 years ago

peilj commented 6 years ago

Description

if I want to predict daily data, how to set the arima parameter m ?

Steps/Code to Reproduce

Expected Results

Actual Results

Versions

tgsmith61591 commented 6 years ago

Depending on the periodicity of your data, it would likely be set to 365. Can't say definitively without knowing your data, though.

Closing issue since it's more of an inquiry.

peilj commented 6 years ago

@tgsmith61591 thank for your reply I have four years history daily data, I want to predict the data for the next 90 days by pyramid. When i used pyramid i don't know how to set the arima parameter m

tgsmith61591 commented 6 years ago

If fitting using auto_arima, you can simply set m=365. If fitting a direct ARIMA model, the parameter is seasonal_order, and it expects a tuple in the form of (P, D, Q, m).

http://pyramid-arima.readthedocs.io/en/latest/_submodules/arima.html#auto-arima

On Dec 1, 2017 2:06 AM, "peilj" notifications@github.com wrote:

@tgsmith61591 https://github.com/tgsmith61591 thank for your reply I have four years history daily data, I want to predict the data for the next 90 days by pyramid. When i used pyramid i don't know how to set the arima parameter m

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/tgsmith61591/pyramid/issues/19#issuecomment-348428798, or mute the thread https://github.com/notifications/unsubscribe-auth/AF10omAh5aU3ujST79iH4vaU1hCnzqzkks5s77ObgaJpZM4QwNLe .

tgsmith61591 commented 6 years ago

Also, potential duplicate of #12. If the docs don't solve your issue, take a look at the issue thread and the linked (in the thread) example ipynb.