10.md

ARIMA defferancing poceudre

• deferancing 1 : take Ynew = Yt - Yt-1
• deferancing 2 : take Yd2 = Ynewt - Ynewt-1

Trend (Lag-1) defferencing

• Lag-1 means difeancing opeartion is Yt - Yt-1
• remove trend
• denoted as ARIMA(p,d,q)
• Order d = {0,1,2,..}
  • d=0 means no deferancing(no trend)
  • d=1 means perform deferancing once (linear trend)
  • d=2 means double-defenacing

Seasonal (Lag-M) defferencing

• Lag-M means differancing operation is Yt - Yt-M

• remove seasonal trends

• for cycle with M seasons use Lag-M

• denoted as ARIMA(p,d,q)(P,D,Q)

• Order D = {0=none, 1=once}

  • D=0 means do not perform seasonal differancing
  • D=1 means perform seasonal differancing ONCE

• [Referance] (https://www.youtube.com/watch?v=0xHf-SJ9Z9U)

Referances

• https://www.youtube.com/watch?v=Prpu_U5tKkE

• https://www.coursera.org/learn/practical-time-series-analysis?action=enroll

• https://www.youtube.com/watch?v=t3VTR_hhxiI&index=16&list=PLtiTxpFJ4k6TZ0g496fVcQpt_-XJRNkbi

• https://blog.statsbot.co/time-series-anomaly-detection-algorithms-1cef5519aef2