You are on page 1of 1

Course: STAT3002 Lab#5-Week 9

Topic: Times Series Dated: March 25, 2024

In Time Series analysis, various models are used to represent different time series data. These
models include:

a. Moving Average Model (MA): Also known as MA(q), this model represents a time series
where each value is a linear combination of past error terms.

b. Autoregressive Model (AR): Also referred to as AR(p), this model represents a time series
where each value is a linear combination of its past values.

c. Autoregressive Moving Average Model (ARMA): Combining elements of both AR and MA


models, ARMA(p,q) represents a time series where each value is a combination of past values
and past error terms.

d. Autoregressive Integrated Moving Average Model (ARIMA): Generally used for non-
stationary time series, ARIMA(p,d,q) includes differencing to achieve stationarity along with
the ARMA(p,q) components.

A. Match the time series represented to any of the above:

I. 𝑦𝑡 − 𝑦𝑡−1 = 𝜗1 (𝑦𝑡−2 − 𝑦𝑡−3 ) + 𝜗3 (𝑦𝑡−4 − 𝑦𝑡−5 ) + ∅1 𝑒𝑡−1 − ∅2 𝑒𝑡−2


II. 𝑦𝑡 = 𝜗1 𝑒𝑡 − 𝜗2 𝑒𝑡−1 − 𝜗3 𝑒𝑡−2
III. 𝑦𝑡 = 𝜑1 𝑦𝑡−2 + 𝑒𝑡−1 + 𝑒𝑡−2
IV. 𝑦𝑡 = 𝜑1 𝑦𝑡−1 − 𝜑2 𝑦𝑡−2 + ∅1 𝑒𝑡−1 − 𝜃2 𝑒𝑡−2
Here, 𝜗1 , 𝜗2 , 𝜗3 , ∅1 , ∅2 , 𝜑1 , 𝜑2 ∈ 𝑅 .

B. We will now identify autoregressive and moving average models using the autocorrelation
function and the partial autocorrelation function. Before we can use these models, the time
series must first be stationary. Some important takeaways:

✓ If the ACF plot shows a gradual decrease and tails off to zero, it suggests a stationary
time series, which is typical of an MA model.
✓ If the PACF plot has significant spikes at the first few lags and then drops off, it suggests
an AR model.
✓ If both the ACF and PACF plots have significant spikes at the first few lags and then tail
off, it suggests an ARMA model.

We will be using the lynx dataset to explore these models.

Procedure:

a. Produce a plot of the ACF and PACF of the dataset.


b. Produce a plot of the ACF and PACF of normally randomly distributed data. How do
the results here and in (a) compare?
c. Based on your ACF AND PACF plots, determine a suitable MA or AR model for the
lynx data. Produce equations of these models and interpretations of the equation.

Prepared by: Joshua Harvey

You might also like