The main limitation of the ts time-series R object class (besides the aforementioned x axis issue) is that it cannot deal with irregular time-series. To overcome this problem, we have several alternatives in R.

The zoo package and its reverse dependent xts packages are ts-compatible classes with tons of extremely useful methods. For a quick example, let's build a zoo object from our data, and see how it's represented by the default plot:

> library(zoo)
> zd <- zoo(daily[, -1, with = FALSE], daily[[1]])
> plot(zd)

As we have defined the date column to act as the timestamp of the observations, it's not shown here. The x axis has a nice human-friendly date annotation, which is really pleasant after having checked a bunch of integer-annotated plots in the previous pages.

Of course, zoo supports most of the ts methods, such as diff, lag or cumulative sums; these can be very useful for visualizing data velocity:

> plot(cumsum(zd))

Here, the linear line for the N variable suggests that we do not have any missing values and our dataset includes exactly one data point per day. On the other hand, the steep elevation of the Cancelled line in February highlights that a single day contributed a lot to the overall number of cancelled flights in 2011.