Scribd
Upload
22 views

Introduction To Time Series Analysis

This document provides an introduction to time series analysis. It defines time series as observations indexed by time and discusses stationary versus non-stationary time series. It explains that stationary time series have the best linear predictor and describes methods for converting non-stationary time series to stationary, including removing trends and seasonality. The document also covers stationary time series prediction models like AR, ARMA and SARIMA and discusses their properties.

Uploaded by

datanalyst.id
Copyright
© © All Rights Reserved
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
22 views

Introduction To Time Series Analysis

This document provides an introduction to time series analysis. It defines time series as observations indexed by time and discusses stationary versus non-stationary time series. It explains that stationary time series have the best linear predictor and describes methods for converting non-stationary time series to stationary, including removing trends and seasonality. The document also covers stationary time series prediction models like AR, ARMA and SARIMA and discusses their properties.

Uploaded by

datanalyst.id
Copyright
© © All Rights Reserved
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 17

Introduction to Time Series

Analysis

InKwan Yu
Time Series?
 A set of observations indexed by time t
 Discrete and continuous time series
Stationary Time Series
 (Weakly) stationary
 The covariance is independent of t for each
h
 X ( X t , X t  h )  E ( X t   )( X t  h   )

 The mean is independent of t


E( X t )  
Why Stationary Time Series?
 Stationary time series have the best
linear predictor.
 Nonstationary time series models are
usually slower to implement for
prediction.
Converting Nonstationary Time
Series to Stationary Time Series
 Remove deterministic factors
 Trends
 Polynomial regression fitting
 Exponential smoothing
 Moving average smoothing
 Differencing (B is a back shift operator)
  X t  X t 1  (1  B ) X t
Converting Nonstationary Time
Series to Stationary Time Series
 Remove deterministic factors
 Seasonality (usually combined with trends
removal)
X t  mt  st  Yt
 Differencing
 d  X t  X t  d  (1  B ) X t
d
Converting Nonstationary Time
Series to Stationary Time Series
 Example
Converting Nonstationary Time
Series to Stationary Time Series
Converting Nonstationary Time
Series to Stationary Time Series
 After conversion, remaining data points
are called residuals
 If residuals are IID, then no more
analysis is necessary since its mean
value will be the best predictor
Wold Decomposition
 Stationary time series can be
represented as the following

X t   j Z t  j  Vt ,
j 0

{Z t } ~ WN (0,  ),
Cov ( Z t , Vt )  0,
Vt : Deterministic,
 j  ,
 2
Stationary Time Series
Prediction
 Pn is a prediction function of Xn+h with
forward lag h from Xn.
Pn X n  h  a0  a1 X n    an X 1

 The prediction error is measured in the


minimum mean square
S (a0 , , an )  E ( X n  h  (a0  a1 X n    an X 1 ))
2
Stationary Time Series
Prediction
 Since S is a quadratic function, the
minimum value will be obtained when
all the partial derivatives are 0.
S (a0 ,  , an )
 0, j  0,  n
a j
Stationary Time Series
Prediction
 In another form
Stationary Models
 AR (AutoRegressive)
X t  1 X t 1     p X 1 p  Z t ,
Z t ~ WN (0,  ),2

 ( X s , Z t )  0, s  t.

 AR’s predictor
Pn X n 1  1 X n     p X n 1 p
Stationary Models
 ARMA
 Reduces large autocovariance functions
 A transformed linear predictor is used
Other Models
 Mutivariate Cointegration
 ARIMA
 SARIMA
 FARIMA
 GARCH
References
 Introduction to Time Series and
Forecasting 2nd ed., P. Brockwell and R.
Davis, Springer Verlag
 Adaptive Filter Theory 4th ed., Simon
Haykin, Prentice Hall
 Time Series Analysis, James Douglas
Hamilton, Princeton University Press

You might also like