RStudio, R, and Time Series

pbenavides

Packages

The tidyverse

library(tidyverse)

── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr     1.1.4     ✔ readr     2.1.5
✔ forcats   1.0.0     ✔ stringr   1.5.1
✔ ggplot2   3.5.2     ✔ tibble    3.3.0
✔ lubridate 1.9.4     ✔ tidyr     1.3.1
✔ purrr     1.1.0     
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors

• tidyverse is a meta-package that loads the core packages of the tidyverse.

We will always load all the required packages a the beginning of the document. When loading the tidyverse, it shows which packages are being attached, as well as any conflicts with previously loaded packages.

Core packages

• dplyr is the core package for data transformation. It is paired up with the following packages for specific column types:
  • stringr for strings.
  • forcats for factors (R’s categorical data type).
  • lubridate for dates and date-times.
• ggplot2 is the primary package for visualization.
• readr is used to import data from delimited files (CSV, TSV, …).
• tibble is a modern reimagining of the data frame, keeping what time has proven to be effective, and throwing out what is not.
• tidyr is used to tidy data, i.e. to ensure that each variable is in its own column, each observation is in its own row, and each value is in its own cell.
• purrr is used for functional programming with R.

The tidyverts

library(fpp3)

Registered S3 method overwritten by 'tsibble':
  method               from 
  as_tibble.grouped_df dplyr

── Attaching packages ──────────────────────────────────────────── fpp3 1.0.1 ──

✔ tsibble     1.1.6     ✔ feasts      0.4.1
✔ tsibbledata 0.4.1     ✔ fable       0.4.1

── Conflicts ───────────────────────────────────────────────── fpp3_conflicts ──
✖ lubridate::date()    masks base::date()
✖ dplyr::filter()      masks stats::filter()
✖ tsibble::intersect() masks base::intersect()
✖ tsibble::interval()  masks lubridate::interval()
✖ dplyr::lag()         masks stats::lag()
✖ tsibble::setdiff()   masks base::setdiff()
✖ tsibble::union()     masks base::union()

• fpp3 is also a meta-package that load the tidyverts ecosystem for time series analysis and forecasting.

The tidyverts packages are made to work seamlessly with the tidyverse.

Note

• tsibble is the main data structure we will use to analyze and model time series. It is a time series tibble.
• feasts provides many functions and tools for feature and statistics extraction for time series.
• fable is the core package for modeling and foreasting time series.

Time Series

tsibble objects

Let’s take a look at tourism in Australia:

tourism

# A tsibble: 24,320 x 5 [1Q]
# Key:       Region, State, Purpose [304]
   Quarter Region   State           Purpose  Trips
     <qtr> <chr>    <chr>           <chr>    <dbl>
 1 1998 Q1 Adelaide South Australia Business  135.
 2 1998 Q2 Adelaide South Australia Business  110.
 3 1998 Q3 Adelaide South Australia Business  166.
 4 1998 Q4 Adelaide South Australia Business  127.
 5 1999 Q1 Adelaide South Australia Business  137.
 6 1999 Q2 Adelaide South Australia Business  200.
 7 1999 Q3 Adelaide South Australia Business  169.
 8 1999 Q4 Adelaide South Australia Business  134.
 9 2000 Q1 Adelaide South Australia Business  154.
10 2000 Q2 Adelaide South Australia Business  169.
# ℹ 24,310 more rows

A tsibble is a modified version of a tibble as to

key_vars(tourism)

[1] "Region"  "State"   "Purpose"

key_data(tourism)

The tsibble has 24320 rows and 5 columns. It shows quarterly data¹ on tourism across Australia. It’s divided by Region, State, and purspose of the trip². How many different states are there?

Australian States

distinct(tourism, State)

Which regions are located in Tasmania?

distinct(filter(tourism, State == "Tasmania"),Region)

Data Transformation: Average trips

To get the average trips by purpose, we need to do the following:

1. Filter the original tsibble to get only the data from East Coast, Tasmania.
2. Convert the data to a tibble.
3. Group by purpose.
4. Summarise by getting the mean of the trips.

With traditional code, this would look something like:

summarise(group_by(as_tibble(filter(tourism, State == "Tasmania", 
                                    Region == "East Coast")), Purpose),
          mean_trips = mean(Trips))

Order of code execution

Note that this code must be read inside-out. This makes it harder to understand, and also harder to debug.

Using the native pipe operator; |>, we can improve the same code:

tourism |>                          
  filter() 

Using the native pipe operator; |>, we can improve the same code:

tourism |>                          
  filter(State == "Tasmania",       
         Region == "East Coast") |> 

Using the native pipe operator; |>, we can improve the same code:

tourism |>                          
  filter(State == "Tasmania",       
         Region == "East Coast") |> 
  as_tibble() |>

Using the native pipe operator; |>, we can improve the same code:

tourism |>                          
  filter(State == "Tasmania",       
         Region == "East Coast") |> 
  as_tibble() |>                    
  group_by() 

Using the native pipe operator; |>, we can improve the same code:

tourism |>                          
  filter(State == "Tasmania",       
         Region == "East Coast") |> 
  as_tibble() |>                    
  group_by(Purpose) |>

Using the native pipe operator; |>, we can improve the same code:

tourism |>                          
  filter(State == "Tasmania",       
         Region == "East Coast") |> 
  as_tibble() |>                    
  group_by(Purpose) |>              
  summarise(mean_trips = )

Using the native pipe operator; |>, we can improve the same code:

tourism |>                          
  filter(State == "Tasmania",       
         Region == "East Coast") |> 
  as_tibble() |>                    
  group_by(Purpose) |>              
  summarise(mean_trips = mean(Trips))

Using the native pipe operator; |>, we can improve the same code:

tourism |>
  filter(State == "Tasmania",
         Region == "East Coast") |>
  as_tibble() |>
  group_by(Purpose) |>
  summarise(mean_trips = mean(Trips))

1

Take the tsibble tourism, then

2

filter by State and Region, then

3

convert to a tibble, then

4

group the tibble by purpose, then

5

summarise by taking the mean trips

The pipe operator |>

The pipe is read as “then”, and it allows us to write code in the order it’s supposed to be run.

It also helps to debug code easier, because you can run each function in order and see where the error is.

TS Visualization

Plotting tourism across time

tourism

Plotting tourism across time

tourism |> 
  filter(State == "Tasmania",
         Region == "East Coast")               

Plotting tourism across time

tourism |> 
  filter(State == "Tasmania",
         Region == "East Coast") |> 
  autoplot(Trips)                

Plotting tourism across time

tourism |> 
  filter(State == "Tasmania",
         Region == "East Coast") |> 
  autoplot(Trips) +
  facet_wrap(vars(Purpose), scale = "free_y")                

Plotting tourism across time

tourism |> 
  filter(State == "Tasmania",
         Region == "East Coast") |> 
  autoplot(Trips) +
  facet_wrap(vars(Purpose), scale = "free_y") +
  theme(legend.position = "none")

1

autoplot() detects the data automatically and proposes a plot accordingly.

2

facet_wrap() Divides a plot into subplots (facets).

3

you can customize endless feautres using theme(). Here, we remove the legend, as it’s redudant.

Time plots

aus_production |> 
  autoplot(Gas)

aus_production |> 
  autoplot(Gas) +
  geom_point()

These are the most basic type of plots. We have the time variable in the \(x\)-axis, and our forecast variable in the \(y\)-axis. Time plots should be line plots, and can include or not points.

Seasonal Plots

aus_production |> 
  gg_season(Gas)

The data here are plotted against a single “season”. It’s useful in identifying years with changes in patterns.

Removing the trend from the data:

Seasonal Subseries Plots

aus_production |> 
  gg_subseries(Gas)

Here we split the plot into many subplots, one for each season. This helps us see clearly the underlying seasonal pattern. The mean for each season is represented as the blue horizontal line.

gg_tsdisplay()

aus_production |> 
  gg_tsdisplay(Gas, plot_type = "season")

This function provides a convenient way to have 3 plots: a time plot, an ACF plot, and a third option that can be customized with one of the following plot types:

plot_type options

• “auto”,
• “partial”,
• “season”,
• “histogram”,
• “scatter”,
• “spectrum”

Exporting data to .csv

tourism |> 
  filter(State == "Tasmania",
         Region == "East Coast") |> 
  mutate(Quarter = as.Date(Quarter)) |> 
  write_csv("./datos/tasmania.csv")

You can export to .csv by providing a tsibble or tibble (or any other type of data frame), by calling write_csv(), and specifying the output file’s name.

Footnotes

1. shown besides the tsibble dimension as [1Q]

2. these are specified in the key argument. This tsibble contains