What does the time index tell us?
Some data are naturally evenly spaced by time. The time series discrete_data shown in the top figure has 20 observations, with one observation appearing at each of the discrete time indices 1 through 20. Discrete time indexing is appropriate for discrete_data.
The time series continuous_series shown in the bottom figure also has 20 observations, it is following the same periodic pattern as discrete_data, but its observations are not evenly spaced. Its first, second, and last observations were observed at times 1.210322, 1.746137, and 20.180524, respectively. Continuous time indexing is natural for continuous_series, however, the observations are approximately evenly spaced, with about 1 observation observed per time unit. Let's investigate using a discrete time indexing for continuous_series.
Cet exercice fait partie du cours
Time Series Analysis in R
Instructions
- Use
plot(___, ___, type = "b")to displaycontinuous_seriesversuscontinuous_time_index, its continuous time index - Create a vector 1:20 to be used as a discrete time index.
- Now use
plot(___, ___, type = "b")to displaycontinuous_seriesversusdiscrete_time_index - Note the various differences between the resulting figures, but the approximation appears reasonable because the overall trend remained preserved
Exercice interactif pratique
Essayez cet exercice en complétant cet exemple de code.
# Plot the continuous_series using continuous time indexing
par(mfrow=c(2,1))
plot(continuous_time_index,___, type = "b")
# Make a discrete time index using 1:20
discrete_time_index <-
# Now plot the continuous_series using discrete time indexing
plot(discrete_time_index,___, type = "b")