Dummy variables for seasonal / cyclical components

Use dummy variables to represent seasonal or cyclical components of your time series!

Scenario: Given the Tinder stats for Crammer Nation University's Delta Apple Pi chapter, interpret bQ2.

In Q2 of every year, it's estimated that Delta Apple Pi gets 329.39 less Tinder matches than Q1 (baseline)!

Just in case you're curious...

  • In Q3, they're estimated to get 63.44 more than Q1 (baseline)
  • In Q4, they're estimated to get 68.53 more than Q1 (baseline)

You may see time series written like this...

y-hat = 1029.23 + 43.89(bYear) - 329.39(bQ2) + 63.44(bQ3) + 68.53(bQ4)

y-hat = T + S

y-hat is the predicted value ("Tinder matches").
T is the trend component of the time series.
S is the seasonal component of the time series.

y-hat = eT x eS

y-hat is the predicted value ("Tinder matches").
T is the trend component of the time series.
S is the seasonal component of the time series.

By using dummy variables... we're adding the seasonal component to the time series regression!

Without them, the regression is only assessing the trend component across the periods of time.

Activate AutoScroll