Week 5 Discussion

Forecasting Methods Time Series Classification of Time Series Data There are four major categories that can be used to classify time series data: trend, seasonal, irregular, and cycle. However, remember, time series data might not have all these characteristics. The following examples will help you identify these categories: • Trend Time series data shows a general pattern of change over the observed years. This is an example of time series data showing a trend. A trend is a gradual movement over a period of years. Some trends are fairly predictable. • Seasonal Time series data shows a repetitive pattern of change within a year. This is an example of time series data showing a seasonal pattern. A seasonal pattern is a cycle that repeats itself during a year. For example, a landscape business may see increased sales during the summer season compared to that in the winter season. • Irregular Time series data shows a random disturbance that follows no pattern. This is an example of time series data showing an irregular pattern. Irregular time series data has no real pattern. In order to make short-term forecasts with irregular time series data, moving averages can be used. • Cycle Time series data shows a repetitive pattern of change around the trend over several years. This is an example of time series data showing a cycle. A cycle occurs over several years and is a movement around a trend. For example, there are peaks and troughs in the business cycle. The economy also moves in cycles.

Two Types of Time Series Models Let’s assume that Y denotes the time series variable, T denotes the trend, S denotes the seasonal component, C denotes the cycle, and I denotes the irregular component. The two types of time series models can then be represented by the following two equations: • Additive Model: Y = T + S + C+ I This equation represents the additive model. This model assumes that all the categories of the time series are independent of one another. Accordingly, it adds the values of the four categories (irregular, seasonal, cycle, and trend) to make a forecast. This model is useful in making predictions about time series data that do not exhibit a trend. This model is adequate in the short run because the values of the four categories do not change much. However, for observations over longer periods of time, the multiplicative model is preferred. • Multiplicative Model: Y = T × S × C × I This equation represents the multiplicative model. This model assumes that there is a multiplicative relationship between the four categories of time series data. Accordingly, it multiplies the values of the four major categories (irregular, seasonal, cycle, and trend) to make a forecast. It is useful in making predictions about time series data that exhibit a trend. © 2017 South University