Download presentation
Presentation is loading. Please wait.
1
Time Series Analysis By Tyler Moore
2
What Is Time Series Data
An ordered sequence of values of a variable at equally spaced time intervals Used for: Obtaining an understanding of underlying forces and structure that produced the observed data Used primarily for forecasting and signal detection and estimation
3
Forecasting: Moving Average
Used to gain better information on trends going on in data Moving average: break down time periods into smaller components Using just the average can be a poor way of modeling future expectations
4
Forecasting: Smoothing
Assigns expontentially smaller weights to older observations. Allows for better analysis of trends Single, double(trends), and triple (trends and seasonality) Ex: triple exponential smoothing Use if data shows trend and seasonality Called the Holt-Winters Method
5
Box-Jenkins Models Combination of Moving Average, and Autoregressive Models Autoregressive model: Linear regression of current value against one or more prior values 3 stages: Model Identification Model Estimation Model Validation
6
Model Identification Assess stationarity and seasonality
7
Model Identification Identify order for autoregressive and moving average terms Autocorrelation or partial autocorrelation plot
8
Example No seasonality Appears stationary
9
Example Continued Values alternate in sign and drop off after lag 2 meaning we use AR(2) model Means we use 2 predictors
10
Example Continued
11
Signal Detection and Estimation
EEG and fMRI data fall under this category
12
References Hamilton, J. D. (1994). Time series analysis (Vol. 2). Princeton: Princeton university press. NIST/SEMATECH e-Handbook of Statistical Methods, 25, 2016.
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.