Scribd
Upload
48 views

Time Series Formula or

Uploaded by

2B Dai Ko DUPLICATE
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
48 views

Time Series Formula or

Uploaded by

2B Dai Ko DUPLICATE
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

1

𝑀𝑒𝑎𝑛 𝐴𝑏𝑠𝑜𝑙𝑢𝑡𝑒 𝐷𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 (𝑴𝑨𝑫) = ∑|𝑌𝑖 − 𝑌̂𝑖 |


𝑛
𝑖

100 𝑌𝑖 − 𝑌̂𝑖
𝑀𝑒𝑎𝑛 𝐴𝑏𝑠𝑜𝑙𝑢𝑡𝑒 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝐸𝑟𝑟𝑜𝑟 (𝑴𝑨𝑷𝑬) = ∑| |
𝑛 𝑌𝑖
𝑖

1 2
𝑀𝑒𝑎𝑛 𝑆𝑞𝑢𝑎𝑟𝑒 𝐸𝑟𝑟𝑜𝑟 (𝑴𝑺𝑬) = ∑(𝑌𝑖 − 𝑌̂𝑖 )
𝑛
𝑖

𝑅𝑜𝑜𝑡 𝑀𝑒𝑎𝑛 𝑆𝑞𝑢𝑎𝑟𝑒 𝐸𝑟𝑟𝑜𝑟 (𝑹𝑴𝑺𝑬) = √𝑀𝑆𝐸

MOVING AVERAGE
𝑌𝑡 + 𝑌𝑡−1 + 𝑌𝑡−𝑘+1
𝐹𝑜𝑟𝑒𝑐𝑎𝑠𝑡 (𝑌̂𝑡+1 ) = , 𝑤ℎ𝑒𝑟𝑒 𝑘 𝑖𝑠 𝑛𝑜 𝑜𝑓 𝑜𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛𝑠
𝑘

WEIGHTED MOVING AVERAGE

𝑌̂𝑡+1 = 𝑤1 𝑌𝑡 + 𝑤2 𝑌𝑡−1 + ⋯ + 𝑤𝑘 𝑌𝑡−𝑘+1


𝑤1 + 𝑤2 + ⋯ + 𝑤𝑘 = 1

EXPONENTIAL SMOOTHING

𝑌̂𝑖+1 = 𝑌̂𝑡 + 𝛼(𝑌𝑡 − 𝑌̂𝑡 )

𝑌̂𝑖+1 = 𝛼𝑌𝑡 + 𝛼(1 − 𝛼)𝑌𝑡−1 + 𝛼(1 − 𝛼)2 𝑌𝑡−2 + ⋯ + 𝛼(1 − 𝛼)𝑛 𝑌𝑡−𝑛 + ⋯
0≤𝛼≤1

STATIONARY DATA WITH ADDITIVE SEASONAL EFFECT

𝐹𝑜𝑟𝑒𝑐𝑎𝑠𝑡 (𝑌̂𝑡+𝑛 ) = 𝐸𝑡 + 𝑆𝑡+𝑛−𝑝

𝑊ℎ𝑒𝑟𝑒, 𝐿𝑒𝑣𝑒𝑙 (𝐸𝑡 ) = 𝛼(𝑌𝑡 − 𝑆𝑡−𝑝 ) + (1 − 𝛼)𝐸𝑡−1

𝑆𝑒𝑎𝑠𝑜𝑛𝑎𝑙 𝐹𝑎𝑐𝑡𝑜𝑟 (𝑆𝑡 ) = 𝛽(𝑌𝑡 − 𝐸𝑡 ) + (1 − 𝛽)𝑆𝑡−𝑝

0 ≤ 𝛼 ≤ 1 𝑎𝑛𝑑 0 ≤ 𝛽 ≤ 1

For initialize the estimate for the first p time periods


𝑝
𝑌𝑖
𝐸𝑡 = ∑ , 𝑡 = 1,2, … , 𝑝
𝑝
𝑖=1

𝑆𝑡 = 𝑌𝑡 − 𝐸𝑡 , 𝑡 = 1,2, … , 𝑝

1
STATIONARY DATA WITH MULTIPLICATIVE SEASONAL EFFECT

𝐹𝑜𝑟𝑒𝑐𝑎𝑠𝑡 (𝑌̂𝑡+𝑛 ) = 𝐸𝑡 ∗ 𝑆𝑡+𝑛−𝑝

Where

𝑌𝑡
𝐿𝑒𝑣𝑒𝑙 (𝐸𝑡 ) = 𝛼 ( ) + (1 − 𝛼)𝐸𝑡−1
𝑆𝑡−𝑝
𝑌𝑡
𝑆𝑒𝑎𝑠𝑜𝑛𝑎𝑙 𝐹𝑎𝑐𝑡𝑜𝑟 (𝑆𝑡 ) = 𝛽 ( ) + (1 − 𝛽)𝑆𝑡−𝑝
𝐸𝑡
0 ≤ 𝛼 ≤ 1 𝑎𝑛𝑑 0 ≤ 𝛽 ≤ 1

For initialize the estimate


𝑝
𝑌𝑖
𝐸𝑡 = ∑ , 𝑡 = 1,2, … , 𝑝
𝑝
𝑖=1

𝑌𝑡
𝑆𝑡 = , 𝑡 = 1,2, … , 𝑝
𝐸𝑡

DOUBLE MOVING AVERAGE


𝑌𝑡 + 𝑌𝑡−1 + ⋯ + 𝑌𝑡−𝑘+1
𝑀𝑡 = ,
𝑘
𝑤ℎ𝑒𝑟𝑒 𝑀𝑡 𝑖𝑠 𝑚𝑜𝑣𝑖𝑛𝑔 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑓𝑜𝑟 𝑡ℎ𝑒 𝑝𝑎𝑠𝑡 𝑘 𝑡𝑖𝑚𝑒 𝑝𝑒𝑟𝑖𝑜𝑑𝑠 𝑖𝑛𝑐𝑙𝑢𝑑𝑖𝑛𝑔 𝑡

𝑀𝑡 + 𝑀𝑡−1 + ⋯ + 𝑀𝑡−𝑘+1
𝐷𝑡 = ,
𝑘
𝑤ℎ𝑒𝑟𝑒 𝐷𝑡 𝑖𝑠 𝑚𝑜𝑣𝑖𝑛𝑔 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑓𝑜𝑟 𝑡ℎ𝑒 𝑝𝑎𝑠𝑡 𝑘 𝑡𝑖𝑚𝑒 𝑝𝑒𝑟𝑖𝑜𝑑𝑠 𝑖𝑛𝑐𝑙𝑢𝑑𝑖𝑛𝑔 𝑡

𝐹𝑜𝑟𝑒𝑐𝑎𝑠𝑡 (𝑌̂𝑡+𝑛 ) = 𝐸𝑡 + 𝑛𝑇𝑡


Where

𝐿𝑒𝑣𝑒𝑙 (𝐸𝑡 ) = 2𝑀𝑡 − 𝐷𝑡


2(𝑀𝑡 − 𝐷𝑡 )
𝑇𝑟𝑒𝑛𝑑 (𝑇𝑡 ) =
(𝑘 − 1)

2
DOUBLE EXPONENTIAL SMOOTHING (HOLT’S METHOD)

𝐹𝑜𝑟𝑒𝑐𝑎𝑠𝑡 (𝑌̂𝑡+𝑛 ) = 𝐸𝑡 + 𝑛𝑇𝑡

Where

𝐿𝑒𝑣𝑒𝑙 (𝐸𝑡 ) = 𝛼𝑌𝑡 + (1 − 𝛼)(𝐸𝑡−1 + 𝑇𝑡−1 )


𝑇𝑟𝑒𝑛𝑑 (𝑇𝑡 ) = 𝛽(𝐸𝑡 − 𝐸𝑡−1 ) + (1 − 𝛽)𝑇𝑡−1
0 ≤ 𝛼 ≤ 1 𝑎𝑛𝑑 0 ≤ 𝛽 ≤ 1
𝑊ℎ𝑒𝑟𝑒 𝑖𝑡 𝑖𝑠 𝑐𝑢𝑠𝑡𝑜𝑚𝑎𝑟𝑦 𝑡𝑜 𝑎𝑠𝑠𝑢𝑚𝑒 𝑡ℎ𝑎𝑡 𝐸1 = 𝑌1 𝑎𝑛𝑑 𝑇1 = 0

HOLT-WINTER’S METHOD FOR ADDITIVE SEASONAL EFFECTS

𝐹𝑜𝑟𝑒𝑐𝑎𝑠𝑡 (𝑌̂𝑡+𝑛 ) = 𝐸𝑡 + 𝑛𝑇𝑡 + 𝑆𝑡+𝑛−𝑝

𝑊ℎ𝑒𝑟𝑒 𝑝 𝑖𝑠 𝑡ℎ𝑒 𝑛𝑜 𝑜𝑓 𝑠𝑒𝑎𝑠𝑜𝑛 𝑖𝑛 𝑡ℎ𝑒 𝑡𝑖𝑚𝑒 𝑠𝑒𝑟𝑖𝑒𝑠; 𝑞𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝑝 = 4, 𝑚𝑜𝑛𝑡ℎ𝑙𝑦 𝑝 = 12


And

𝐿𝑒𝑣𝑒𝑙 (𝐸𝑡 ) = 𝛼(𝑌𝑡 − 𝑆𝑡−𝑝 ) + (1 − 𝛼)(𝐸𝑡−1 + 𝑇𝑡−1 )

𝑇𝑟𝑒𝑛𝑑 (𝑇𝑡 ) = 𝛽(𝐸𝑡 − 𝐸𝑡−1 ) + (1 − 𝛽)𝑇𝑡−1


𝑆𝑒𝑎𝑠𝑜𝑛𝑎𝑙 𝐹𝑎𝑐𝑡𝑜𝑟 (𝑆𝑡 ) = 𝛾(𝑌𝑡 − 𝐸𝑡 ) + (1 − 𝛾)𝑆𝑡−𝑝

0 ≤ 𝛼 ≤ 1 𝑎𝑛𝑑 0 ≤ 𝛽 ≤ 1 𝑎𝑛𝑑 0 ≤ 𝛾 ≤ 1

𝐹𝑜𝑟 𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒


𝑝
𝑌𝑡
𝑆𝑡 = 𝑌𝑡 − ∑ , 𝑡 = 1,2, … , 𝑝
𝑝
𝑖=1

𝐹𝑖𝑟𝑠𝑡 𝐸𝑡 𝑣𝑎𝑙𝑢𝑒 𝑐𝑎𝑛 𝑏𝑒 𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒𝑑 𝑖𝑠 𝑎𝑡 𝑡𝑖𝑚𝑒 𝑝𝑒𝑟𝑖𝑜𝑑 𝑡 = 𝑝 + 1 𝑤ℎ𝑒𝑟𝑒 𝑆𝑡−𝑝 𝑖𝑠 𝑘𝑛𝑜𝑤𝑛

𝐹𝑜𝑟 𝑎𝑙𝑙 𝑜𝑡ℎ𝑒𝑟 𝑣𝑎𝑙𝑢𝑒𝑠:


𝐸𝑝 = 𝑌𝑝 − 𝑆𝑝

𝐸𝑝−1 = 𝐸𝑝−2 = ⋯ = 𝐸1 = 0

𝑇𝑝 = 𝑇𝑝−1 = ⋯ = 𝑇1 = 0

𝑌̂𝑝 = 𝑌̂𝑝−1 = ⋯ = 𝑌̂1 = 0

3
HOLT-WINTER’S METHOD FOR MULTIPLICATIVE SEASONAL EFFECTS

𝐹𝑜𝑟𝑒𝑐𝑎𝑠𝑡 (𝑌̂𝑡+𝑛 ) = (𝐸𝑡 + 𝑛𝑇𝑡 ) ∗ 𝑆𝑡+𝑛−𝑝

𝑊ℎ𝑒𝑟𝑒 𝑝 𝑖𝑠 𝑡ℎ𝑒 𝑛𝑜 𝑜𝑓 𝑠𝑒𝑎𝑠𝑜𝑛 𝑖𝑛 𝑡ℎ𝑒 𝑡𝑖𝑚𝑒 𝑠𝑒𝑟𝑖𝑒𝑠; 𝑞𝑢𝑎𝑟𝑡𝑒𝑟𝑙𝑦 𝑝 = 4, 𝑚𝑜𝑛𝑡ℎ𝑙𝑦 𝑝 = 12


And

𝑌𝑡
𝐿𝑒𝑣𝑒𝑙 (𝐸𝑡 ) = 𝛼 ( ) + (1 − 𝛼)(𝐸𝑡−1 + 𝑇𝑡−1 )
𝑆𝑡−𝑝

𝑇𝑟𝑒𝑛𝑑 (𝑇𝑡 ) = 𝛽(𝐸𝑡 − 𝐸𝑡−1 ) + (1 − 𝛽)𝑇𝑡−1


𝑌𝑡
𝑆𝑒𝑎𝑠𝑜𝑛𝑎𝑙 𝐹𝑎𝑐𝑡𝑜𝑟 (𝑆𝑡 ) = 𝛾 ( ) + (1 − 𝛾)𝑆𝑡−𝑝
𝐸𝑡
0 ≤ 𝛼 ≤ 1 𝑎𝑛𝑑 0 ≤ 𝛽 ≤ 1 𝑎𝑛𝑑 0 ≤ 𝛾 ≤ 1

𝐹𝑜𝑟 𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒


𝑌𝑡
𝑆𝑡 = , 𝑡 = 1,2, … , 𝑝
𝑌𝑡
∑𝑝𝑖=1
𝑝

𝐹𝑖𝑟𝑠𝑡 𝐸𝑡 𝑣𝑎𝑙𝑢𝑒 𝑐𝑎𝑛 𝑏𝑒 𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒𝑑 𝑖𝑠 𝑎𝑡 𝑡𝑖𝑚𝑒 𝑝𝑒𝑟𝑖𝑜𝑑 𝑡 = 𝑝 + 1 𝑤ℎ𝑒𝑟𝑒 𝑆𝑡−𝑝 𝑖𝑠 𝑘𝑛𝑜𝑤𝑛

𝐹𝑜𝑟 𝑎𝑙𝑙 𝑜𝑡ℎ𝑒𝑟 𝑣𝑎𝑙𝑢𝑒𝑠:


𝑌
𝐸𝑝 = 𝑆𝑝
𝑝

𝐸𝑝−1 = 𝐸𝑝−2 = ⋯ = 𝐸1 = 0

𝑇𝑝 = 𝑇𝑝−1 = ⋯ = 𝑇1 = 0

𝑌̂𝑝 = 𝑌̂𝑝−1 = ⋯ = 𝑌̂1 = 0

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy