4.1.1 Basic Time Series Plots

The easiest way is to use autoplot from the feasts package. You may need to specify the actual measured variable, if there is more than one numerical column:

The R package timetk gives us interactive plots that may be more evocative than the static plot above. The basic plot function with timetk is plot_time_series. There are arguments for the date variable, the value you want to plot, colours, groupings etc.

Let us explore this dataset using timetk, using our trusted method of asking Questions:

Q.1 How are the weekly sales different for each Department?

number of Departments. So we should be fine plotting them and also facetting with them, as we will see in a bit:

Q.2. What do the sales per Dept look like during the month of December (Christmas time) in 2012? Show the individual Depts as facets.

We can of course zoom into the interactive plot above, but if we were to plot it anyway:

Clearly the “unfortunate” Dept#13 has seen something of a Christmas drop in sales, as has Dept#38 ! The rest, all is well, it seems…

Too much noise? How about some averaging?

Q.3 How do we smooth out some of the variations in the time series to be able to understand it better?

Sometimes there is too much noise in the time series observations and we want to take what is called a rolling average. For this we will use the function timetk::slidify to create an averaging function of our choice, and then apply it to the time series using regular dplyr::mutate. Let’s take the average of Sales for each month in each Department. Our function will be named “rolling_avg_month”:

OK, slidify creates a function! Let’s apply it to the Walmart Sales time series…

Graphs are smoother now. Need to check whether the averaging was done on a per-Dept basis…should we have had a group_by(Dept) before the averaging, and ungroup() before plotting? Try it !!

4.1.2 Decomposing Time Series: Trends, Seasonal Patterns, and Cycles

Each data point (Yt) at time t in a Time Series can be expressed as either a sum or a product of 4 components, namely, Seasonality(St), Trend(Tt), Cyclic, and Error(et) (a.k.a White Noise).

• Trend: pattern exists when there is a long-term increase or decrease in the data.
• Seasonal: pattern exists when a series is influenced by seasonal factors (e.g., the quarter of the year, the month, or day of the week).
• Cyclic: pattern exists when data exhibit rises and falls that are not of fixed period (duration usually of at least 2 years).
• Error or Noise: Random component

Decomposing non-seasonal data means breaking it up into trend and irregular components. To estimate the trend component of a non-seasonal time series that can be described using an additive model, it is common to use a smoothing method, such as calculating the simple moving average of the time series.

timetk has the ability to achieve this: Let us plot the trend, seasonal, cyclic and irregular aspects of Weekly_Sales for Dept 38:

We can do this for all Dept using fable and fabletools:

4.1.3 Case Study-2: Dataset from nycflights13

Let us try the flights dataset from the package nycflights13. Try data(package = "nycflights13") in your Console.

We have the following datasets in nycflights13:

• airlines Airline names.
• airports Airport metadata
• flights Flights data
• planes Plane metadata.
• weather Hourly weather data

Let us analyze the flights data:

## Rows: 336,776
## Columns: 19
## $ year           <int> 2013, 2013, 2013, 2013, 2013, 2013, 2013, 2013, 2013, 2…
## $ month          <int> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1…
## $ day            <int> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1…
## $ dep_time       <int> 517, 533, 542, 544, 554, 554, 555, 557, 557, 558, 558, …
## $ sched_dep_time <int> 515, 529, 540, 545, 600, 558, 600, 600, 600, 600, 600, …
## $ dep_delay      <dbl> 2, 4, 2, -1, -6, -4, -5, -3, -3, -2, -2, -2, -2, -2, -1…
## $ arr_time       <int> 830, 850, 923, 1004, 812, 740, 913, 709, 838, 753, 849,…
## $ sched_arr_time <int> 819, 830, 850, 1022, 837, 728, 854, 723, 846, 745, 851,…
## $ arr_delay      <dbl> 11, 20, 33, -18, -25, 12, 19, -14, -8, 8, -2, -3, 7, -1…
## $ carrier        <chr> "UA", "UA", "AA", "B6", "DL", "UA", "B6", "EV", "B6", "…
## $ flight         <int> 1545, 1714, 1141, 725, 461, 1696, 507, 5708, 79, 301, 4…
## $ tailnum        <chr> "N14228", "N24211", "N619AA", "N804JB", "N668DN", "N394…
## $ origin         <chr> "EWR", "LGA", "JFK", "JFK", "LGA", "EWR", "EWR", "LGA",…
## $ dest           <chr> "IAH", "IAH", "MIA", "BQN", "ATL", "ORD", "FLL", "IAD",…
## $ air_time       <dbl> 227, 227, 160, 183, 116, 150, 158, 53, 140, 138, 149, 1…
## $ distance       <dbl> 1400, 1416, 1089, 1576, 762, 719, 1065, 229, 944, 733, …
## $ hour           <dbl> 5, 5, 5, 5, 6, 5, 6, 6, 6, 6, 6, 6, 6, 6, 6, 5, 6, 6, 6…
## $ minute         <dbl> 15, 29, 40, 45, 0, 58, 0, 0, 0, 0, 0, 0, 0, 0, 0, 59, 0…
## $ time_hour      <dttm> 2013-01-01 05:00:00, 2013-01-01 05:00:00, 2013-01-01 0…

We have time-related columns; Apart from year, month, day we have time_hour; and time-event numerical data such as arr_delay (arrival delay) and dep_delay (departure delay). We also have categorical data such as carrier, origin, dest, flight and tailnum of the aircraft. It is also a large dataset containing 330K entries. Enough to play with!!

Let us replace the NAs in arr_delay and dep_delay with zeroes for now, and convert it into a time-series object with tsibble:

Let us proceed with our questions:

Q.1. Plot the monthly average arrival delay by carrier

Q.2. Plot a candlestick chart for total flight delays by month for each carrier