Time Series Plots

class: center, middle, title-slide

.title[
# Time Series Plots
]
.author[
### Luke Tierney
]
.institute[
### University of Iowa
]
.date[
### 2025-04-09
]

---

layout: true

## Time Series Data
---

A lot of data is collected over time:
--

- economic performace measures;

- weather, climate data;

- medical information.

The objectives can be

- understanding the past &ndash; how we got where we are;

- understanding patterns in variation;

- forecasting what will happen next.

---

Even if time is not of primary interest, data are often collected with
a time stamp.

Examining the behavior over time can reveal interesting features that
may hint at confounding variables:

- temperature varying with time of day;

- humidity varying over the year.

---

Time series data can be collected at regularly spaced intervals:

- yearly, e.g. the `gapminder` data

- monthly, e.g. the `river` data

- daily, e.g. high temperatures.

Data might be missing

- irregularly, e.g. faulty instrument, holidays;

- regularly, e.g. weekends.

Time series can also be recorded at irregular times:

- blood pressure at doctor visits;

Variables recorded over time can be numerical or categorical.

---
layout: false
## Time Series Objects

Time series data can be represented as regular data frames with a time
variable.

There are also a number of specialized object classes for dealing with time
series.

A number of packages provide `plot` or `autoplot` methods and other
utilities for these objects.

Some of the most useful:

| Class      |  Package   | Features                                           |
| ---------- | ---------- | -------------------------------------------------- |
|`ts`, `mts`| `base`     | regular series; methods for `plot` and `lines`.   |
|           | `forecast` | methods for `autoplot` and `autolayer`.          |
|`zoo`      |  `zoo`     | irregular and regular series; `autoplot` support. |
|           | `broom`    |`tidy` for converting to tidy data frames.         |

---

## `autoplot` and `autolayer`

Many data structures provide methods for `autoplot` and `autolayer`.

From the `autoplot` help page:

> `autoplot` uses `ggplot2` to draw a particular plot for an object of
> a particular class in a single command. This defines the S3 generic
> that other classes and packages can extend.

From the `autolayer` help page:

> `autolayer` uses `ggplot2` to draw a particular layer for an object
> of a particular class in a single command. This defines the S3
> generic that other classes and packages can extend.

---
layout: true
## Example: River Data
---

Creating a `ts` object for the `river` data:

``` r
river <- scan("https://www.stat.uiowa.edu/~luke/data/river.dat")
(riverTS <- ts(river))
## Time Series:
## Start = 1 
## End = 128 
## Frequency = 1 
## [1] 8.95361 9.49409 10.19430 10.95660 11.07770 10.98170 10.48970 10.27540
## [9] 10.16620 9.23484 8.56537 8.51064
## [ reached getOption("max.print") -- omitted 116 entries ]
```
--

Or, since the data were collected monthly:

``` r
(riverTS <- ts(river, frequency = 12))
## Jan Feb Mar Apr May Jun Jul Aug
## 1 8.95361 9.49409 10.19430 10.95660 11.07770 10.98170 10.48970 10.27540
## Sep Oct Nov Dec
## 1 10.16620 9.23484 8.56537 8.51064
## [ reached getOption("max.print") -- omitted 10 rows ]
```

---

Using `tidy` from package `broom` to create a data frame from the `ts` object:

``` r
riverDF <- broom::tidy(riverTS)
head(riverDF, 3)
## # A tibble: 3 × 2
## index value
## <dbl> <dbl>
## 1 1 8.95
## 2 1.08 9.49
## 3 1.17 10.2
```
--

Change to more convenient names:

``` r
riverDF <- rename(riverDF, time = index, flow = value)
head(riverDF, 3)
## # A tibble: 3 × 2
## time flow
## <dbl> <dbl>
## 1 1 8.95
## 2 1.08 9.49
## 3 1.17 10.2
```

---
layout: true
## Basic Plots for a Single Numeric Time Series
---

Numeric time series are usually plotted as a _line chart_.

This is what the `autoplot` method for `ts` objects provided by
package `forecast` does.

For the river flow data:

.pull-left[
<img src="timeseries_files/figure-html/river-ts-autoplot-1.png" style="display: block; margin: auto;" />
]

.pull-right[

``` r
library(forecast)
p <- autoplot(riverTS)
p
```
]

---

Using symbols in addition to lines is sometimes useful:

.pull-left[
<img src="timeseries_files/figure-html/river-ts-autoplot-syms-1.png" style="display: block; margin: auto;" />
]

.pull-right[

``` r
p + geom_point()
```
]

---
layout: true
## Aspect Ratio
---

Aspect ratio has a significant effect on how well we perceive slope
differences.

- Banking to 45 degrees: choose an aspect ratio so the magnitude of
  the important slopes is close to 45 degrees.

- This usually requires the horizontal axis to be longer than the
  vertical axis.

---

Controlling the aspect ratio:

- For `base` graphics use the `asp` argument to `plot`.

- `lattice` uses the `aspect` argument.

- For `ggplot` use the `ratio` argument to `coord_fixed`.

``` r
p + coord_fixed(ratio = 1 / 3)
```

<img src="timeseries_files/figure-html/unnamed-chunk-7-1.png" style="display: block; margin: auto;" />
---

Lattice supports several aspect ratio strategies:

- `"fill"` to avoid unused white space;

- `"xy"` to use a 45 degree banking rule;

- `"iso"` for isometric scales, e.g. distances.

Some work may be needed to avoid excess white space around figures.

Using multiple panels can help.

In an interactive setting, zoom and pan support can help.

---

Multiple panels:

.pull-left[
<img src="timeseries_files/figure-html/river-ts-panels-1.png" style="display: block; margin: auto;" />
]

.pull-right[

``` r
mutate(riverDF,
       period = cut(time, breaks = 3)) |>
    ggplot(aes(time, flow)) +
    geom_line() +
    facet_wrap(~ period,
               scales = "free_x",
               ncol = 1)
```
]

---

Adding zoom and pan support using [`plotly`](https://plotly.com/r/):

``` r
library(plotly)
pp <- p + geom_line() + coord_fixed(ratio = 1 / 3)
ggplotly(pp)
```

<div class="plotly html-widget html-fill-item" id="htmlwidget-8d32cfd9db97395c4a5f" style="width:720px;height:288px;"></div>
<script type="application/json" data-for="htmlwidget-8d32cfd9db97395c4a5f">{"x":{"data":[{"x":[1,1.0833333333333333,1.1666666666666667,1.25,1.3333333333333335,1.4166666666666667,1.5,1.5833333333333335,1.6666666666666667,1.75,1.8333333333333335,1.9166666666666667,2,2.0833333333333335,2.166666666666667,2.25,2.3333333333333335,2.416666666666667,2.5,2.5833333333333335,2.666666666666667,2.75,2.8333333333333335,2.916666666666667,3,3.0833333333333335,3.166666666666667,3.2500000000000004,3.3333333333333335,3.416666666666667,3.5000000000000004,3.5833333333333335,3.666666666666667,3.7500000000000004,3.8333333333333335,3.916666666666667,4,4.0833333333333339,4.166666666666667,4.25,4.3333333333333339,4.416666666666667,4.5,4.5833333333333339,4.666666666666667,4.75,4.8333333333333339,4.916666666666667,5,5.0833333333333339,5.166666666666667,5.2500000000000009,5.3333333333333339,5.416666666666667,5.5000000000000009,5.5833333333333339,5.666666666666667,5.7500000000000009,5.8333333333333339,5.916666666666667,6,6.0833333333333339,6.166666666666667,6.2500000000000009,6.3333333333333339,6.416666666666667,6.5000000000000009,6.5833333333333339,6.666666666666667,6.7500000000000009,6.8333333333333339,6.916666666666667,7,7.0833333333333339,7.166666666666667,7.2500000000000009,7.3333333333333339,7.416666666666667,7.5000000000000009,7.5833333333333339,7.6666666666666679,7.7500000000000009,7.8333333333333339,7.9166666666666679,8,8.0833333333333339,8.1666666666666679,8.25,8.3333333333333339,8.4166666666666679,8.5,8.5833333333333339,8.6666666666666679,8.75,8.8333333333333339,8.9166666666666679,9,9.0833333333333339,9.1666666666666679,9.2500000000000018,9.3333333333333339,9.4166666666666679,9.5000000000000018,9.5833333333333339,9.6666666666666679,9.7500000000000018,9.8333333333333339,9.9166666666666679,10,10.083333333333334,10.166666666666668,10.250000000000002,10.333333333333334,10.416666666666668,10.500000000000002,10.583333333333334,10.666666666666668,10.750000000000002,10.833333333333334,10.916666666666668,11,11.083333333333334,11.166666666666668,11.250000000000002,11.333333333333334,11.416666666666668,11.500000000000002,11.583333333333334],"y":[8.9536099999999994,9.4940899999999999,10.1943,10.9566,11.0777,10.9817,10.489699999999999,10.275399999999999,10.1662,9.2348400000000002,8.5653699999999997,8.5106400000000004,10.1211,11.184900000000001,10.9352,11.180300000000001,10.9572,10.568300000000001,10.1404,9.9954199999999993,9.1418499999999998,8.6115100000000009,8.3889200000000006,8.3879900000000003,9.9749800000000004,10.268000000000001,11.051,10.517099999999999,10.969099999999999,10.4305,10.410399999999999,10.079700000000001,9.5892300000000006,9.4086099999999995,8.5871399999999998,8.5324100000000005,8.4013799999999996,8.2953799999999998,9.6960499999999996,11.4099,11.2158,10.417,10.046900000000001,10.3744,10.134499999999999,9.1421299999999999,8.69693,9.0465999999999998,9.8736800000000002,11.419499999999999,10.782,11.207100000000001,10.273400000000001,10.2904,9.4990199999999998,9.6448400000000003,9.2041900000000005,8.7909600000000001,9.0131499999999996,9.2212800000000001,9.5860800000000008,9.8040400000000005,10.515599999999999,10.023999999999999,9.9657900000000001,10.7536,10.2445,10.095700000000001,9.3805899999999998,8.8368400000000005,8.8799499999999991,9.7058700000000009,10.474399999999999,11.503,11.589399999999999,10.6264,10.6584,10.540100000000001,10.2864,10.0068,9.2973700000000008,8.7810400000000008,8.8345900000000004,8.9115300000000008,10.1838,10.641,10.033099999999999,9.8803199999999993,11.277699999999999,10.428599999999999,9.7869499999999992,9.4578600000000002,8.8108599999999999,8.6505399999999995,8.80213,9.1636000000000006,9.3950800000000001,10.460000000000001,11.100300000000001,11.167299999999999,10.491400000000001,10.0914,10.0245,9.5253899999999998,9.0230099999999993,8.7327200000000005,8.8243899999999993,9.0724900000000002,10.436999999999999,10.3871,10.8521,10.913,9.9356399999999994,10.263299999999999,10.0245,9.4510299999999994,8.8103599999999993,8.6156699999999997,8.8915100000000002,9.7064800000000009,9.5980299999999996,9.7055000000000007,9.6376100000000005,10.681800000000001,10.6586,10.700799999999999,10.416700000000001,10.1996],"text":["x: 1.000000 y: 8.95361","x: 1.083333 y: 9.49409","x: 1.166667 y: 10.19430","x: 1.250000 y: 10.95660","x: 1.333333 y: 11.07770","x: 1.416667 y: 10.98170","x: 1.500000 y: 10.48970","x: 1.583333 y: 10.27540","x: 1.666667 y: 10.16620","x: 1.750000 y: 9.23484","x: 1.833333 y: 8.56537","x: 1.916667 y: 8.51064","x: 2.000000 y: 10.12110","x: 2.083333 y: 11.18490","x: 2.166667 y: 10.93520","x: 2.250000 y: 11.18030","x: 2.333333 y: 10.95720","x: 2.416667 y: 10.56830","x: 2.500000 y: 10.14040","x: 2.583333 y: 9.99542","x: 2.666667 y: 9.14185","x: 2.750000 y: 8.61151","x: 2.833333 y: 8.38892","x: 2.916667 y: 8.38799","x: 3.000000 y: 9.97498","x: 3.083333 y: 10.26800","x: 3.166667 y: 11.05100","x: 3.250000 y: 10.51710","x: 3.333333 y: 10.96910","x: 3.416667 y: 10.43050","x: 3.500000 y: 10.41040","x: 3.583333 y: 10.07970","x: 3.666667 y: 9.58923","x: 3.750000 y: 9.40861","x: 3.833333 y: 8.58714","x: 3.916667 y: 8.53241","x: 4.000000 y: 8.40138","x: 4.083333 y: 8.29538","x: 4.166667 y: 9.69605","x: 4.250000 y: 11.40990","x: 4.333333 y: 11.21580","x: 4.416667 y: 10.41700","x: 4.500000 y: 10.04690","x: 4.583333 y: 10.37440","x: 4.666667 y: 10.13450","x: 4.750000 y: 9.14213","x: 4.833333 y: 8.69693","x: 4.916667 y: 9.04660","x: 5.000000 y: 9.87368","x: 5.083333 y: 11.41950","x: 5.166667 y: 10.78200","x: 5.250000 y: 11.20710","x: 5.333333 y: 10.27340","x: 5.416667 y: 10.29040","x: 5.500000 y: 9.49902","x: 5.583333 y: 9.64484","x: 5.666667 y: 9.20419","x: 5.750000 y: 8.79096","x: 5.833333 y: 9.01315","x: 5.916667 y: 9.22128","x: 6.000000 y: 9.58608","x: 6.083333 y: 9.80404","x: 6.166667 y: 10.51560","x: 6.250000 y: 10.02400","x: 6.333333 y: 9.96579","x: 6.416667 y: 10.75360","x: 6.500000 y: 10.24450","x: 6.583333 y: 10.09570","x: 6.666667 y: 9.38059","x: 6.750000 y: 8.83684","x: 6.833333 y: 8.87995","x: 6.916667 y: 9.70587","x: 7.000000 y: 10.47440","x: 7.083333 y: 11.50300","x: 7.166667 y: 11.58940","x: 7.250000 y: 10.62640","x: 7.333333 y: 10.65840","x: 7.416667 y: 10.54010","x: 7.500000 y: 10.28640","x: 7.583333 y: 10.00680","x: 7.666667 y: 9.29737","x: 7.750000 y: 8.78104","x: 7.833333 y: 8.83459","x: 7.916667 y: 8.91153","x: 8.000000 y: 10.18380","x: 8.083333 y: 10.64100","x: 8.166667 y: 10.03310","x: 8.250000 y: 9.88032","x: 8.333333 y: 11.27770","x: 8.416667 y: 10.42860","x: 8.500000 y: 9.78695","x: 8.583333 y: 9.45786","x: 8.666667 y: 8.81086","x: 8.750000 y: 8.65054","x: 8.833333 y: 8.80213","x: 8.916667 y: 9.16360","x: 9.000000 y: 9.39508","x: 9.083333 y: 10.46000","x: 9.166667 y: 11.10030","x: 9.250000 y: 11.16730","x: 9.333333 y: 10.49140","x: 9.416667 y: 10.09140","x: 9.500000 y: 10.02450","x: 9.583333 y: 9.52539","x: 9.666667 y: 9.02301","x: 9.750000 y: 8.73272","x: 9.833333 y: 8.82439","x: 9.916667 y: 9.07249","x: 10.000000 y: 10.43700","x: 10.083333 y: 10.38710","x: 10.166667 y: 10.85210","x: 10.250000 y: 10.91300","x: 10.333333 y: 9.93564","x: 10.416667 y: 10.26330","x: 10.500000 y: 10.02450","x: 10.583333 y: 9.45103","x: 10.666667 y: 8.81036","x: 10.750000 y: 8.61567","x: 10.833333 y: 8.89151","x: 10.916667 y: 9.70648","x: 11.000000 y: 9.59803","x: 11.083333 y: 9.70550","x: 11.166667 y: 9.63761","x: 11.250000 y: 10.68180","x: 11.333333 y: 10.65860","x: 11.416667 y: 10.70080","x: 11.500000 y: 10.41670","x: 11.583333 y: 10.19960"],"type":"scatter","mode":"lines","line":{"width":1.8897637795275593,"color":"rgba(0,0,0,1)","dash":"solid"},"hoveron":"points","showlegend":false,"xaxis":"x","yaxis":"y","hoverinfo":"text","frame":null}],"layout":{"margin":{"t":34.264840182648406,"r":7.3059360730593621,"b":60.174346201743461,"l":49.215442092154419},"font":{"color":"rgba(0,0,0,1)","family":"","size":21.253632212536321},"xaxis":{"domain":[0,1],"automargin":true,"type":"linear","autorange":false,"range":[0.47083333333333333,12.112500000000001],"tickmode":"array","ticktext":["2","4","6","8","10","12"],"tickvals":[2,3.9999999999999996,6,8,10,12],"categoryorder":"array","categoryarray":["2","4","6","8","10","12"],"nticks":null,"ticks":"","tickcolor":null,"ticklen":3.6529680365296811,"tickwidth":0,"showticklabels":true,"tickfont":{"color":"rgba(77,77,77,1)","family":"","size":17.002905770029059},"tickangle":-0,"showline":false,"linecolor":null,"linewidth":0,"showgrid":true,"gridcolor":"rgba(235,235,235,1)","gridwidth":0.66417600664176002,"zeroline":false,"anchor":"y","title":{"text":"Time","font":{"color":"rgba(0,0,0,1)","family":"","size":21.253632212536321}},"scaleanchor":"y","scaleratio":3,"hoverformat":".2f"},"yaxis":{"domain":[0,1],"automargin":true,"type":"linear","autorange":false,"range":[8.1306790000000007,11.754100999999999],"tickmode":"array","ticktext":["9","10","11"],"tickvals":[9,10,11],"categoryorder":"array","categoryarray":["9","10","11"],"nticks":null,"ticks":"","tickcolor":null,"ticklen":3.6529680365296811,"tickwidth":0,"showticklabels":true,"tickfont":{"color":"rgba(77,77,77,1)","family":"","size":17.002905770029059},"tickangle":-0,"showline":false,"linecolor":null,"linewidth":0,"showgrid":true,"gridcolor":"rgba(235,235,235,1)","gridwidth":0.66417600664176002,"zeroline":false,"anchor":"x","title":{"text":"riverTS","font":{"color":"rgba(0,0,0,1)","family":"","size":21.253632212536321}},"scaleanchor":"x","scaleratio":0.33333333333333331,"hoverformat":".2f"},"shapes":[{"type":"rect","fillcolor":"transparent","line":{"color":"rgba(77,77,77,1)","width":0.66417600664176002,"linetype":"solid"},"yref":"paper","xref":"paper","x0":0,"x1":1,"y0":0,"y1":1}],"showlegend":false,"legend":{"bgcolor":null,"bordercolor":null,"borderwidth":0,"font":{"color":"rgba(0,0,0,1)","family":"","size":17.002905770029059}},"hovermode":"closest","barmode":"relative"},"config":{"doubleClick":"reset","modeBarButtonsToAdd":["hoverclosest","hovercompare"],"showSendToCloud":false},"source":"A","attrs":{"11a3947643e13":{"x":{},"y":{},"type":"scatter"},"11a3926010f9f":{"x":{},"y":{}}},"cur_data":"11a3947643e13","visdat":{"11a3947643e13":["function (y) ","x"],"11a3926010f9f":["function (y) ","x"]},"highlight":{"on":"plotly_click","persistent":false,"dynamic":false,"selectize":false,"opacityDim":0.20000000000000001,"selected":{"opacity":1},"debounce":0},"shinyEvents":["plotly_hover","plotly_click","plotly_selected","plotly_relayout","plotly_brushed","plotly_brushing","plotly_clickannotation","plotly_doubleclick","plotly_deselect","plotly_afterplot","plotly_sunburstclick"],"base_url":"https://plot.ly"},"evals":[],"jsHooks":[]}</script>

---

There is not always a single best aspect ratio.

The `co2` data set in the `datasets` package contains monthly
concentrations of CO2 for the years 1959 to 1997 recorded at the Mauna
Loa observatory.

The `co2` data is stored as an object of class `ts`:

``` r
str(co2)
##  Time-Series [1:468] from 1959 to 1998: 315 316 316 318 318 ...
```
--

`plot` and `xyplot` have methods for `ts` objects that simplify time
series plotting.

---

Two aspect ratios:

.pull-left[

``` r
xyplot(co2)
```

<img src="timeseries_files/figure-html/unnamed-chunk-11-1.png" style="display: block; margin: auto;" />
]

.pull-right[

``` r
xyplot(co2, aspect = "xy")
```

<img src="timeseries_files/figure-html/unnamed-chunk-12-1.png" style="display: block; margin: auto;" />
Is there a slowdown in the increase?

A graph with more recent data is
[available](https://gml.noaa.gov/ccgg/trends/).

The full data is
[available](https://gml.noaa.gov/webdata/ccgg/trends/co2/co2_annmean_mlo.txt)
as well.
]

---

`autoplot` does not yet provide an easy way to bank to 45 degrees; you
need to experiment or calculate an appropriate aspect ratio yourself.

The default `autoplot`:

.pull-left[

<img src="timeseries_files/figure-html/co2-autoplot-default-1.png" style="display: block; margin: auto;" />
]

.pull-right[

``` r
library(forecast)
autoplot(co2)
```
]

---

Aspect ratio adjusted with `coord_fixed`:

``` r
r45 <- median(diff(time(co2))) / median(abs(diff(co2)))
autoplot(co2) + coord_fixed(ratio = r45)
```

---
layout: true
## Updated Mauna Loa CO2 Data
---

Downloading and reading in the data:

``` r
library(lubridate)
url <- "https://gml.noaa.gov/webdata/ccgg/trends/co2/co2_mm_mlo.txt"
if (! file.exists("co2new.dat") ||
 now() > file.mtime("co2new.dat") + weeks(4))
 download.file(url, "co2new.dat")
co2new <- read.table("co2new.dat")
```

Clean up the variable names:

``` r
head(co2new)
## V1 V2 V3 V4 V5 V6 V7 V8
## 1 1958 3 1958.203 315.71 314.44 -1 -9.99 -0.99
## 2 1958 4 1958.288 317.45 315.16 -1 -9.99 -0.99
## 3 1958 5 1958.370 317.51 314.69 -1 -9.99 -0.99
## 4 1958 6 1958.455 317.27 315.15 -1 -9.99 -0.99
## 5 1958 7 1958.537 315.87 315.20 -1 -9.99 -0.99
## 6 1958 8 1958.622 314.93 316.21 -1 -9.99 -0.99
names(co2new) <- c("year", "month", "decimal_date", "average",
 "deseason", "ndays", "stdd", "umm")
```

---

An initial look:

``` r
ggplot(co2new, aes(decimal_date, average)) + geom_line()
```

---

Missing values are encoded as `-9.99` or `-99.99`.

If there are any such values they need to be replaced with R's missing
value representation:

``` r
filter(co2new, average < -9)
## [1] year month decimal_date average deseason 
## [6] ndays stdd umm 
## <0 rows> (or 0-length row.names)
co2new <- mutate(co2new, average = ifelse(average < -9, NA, average))
```

---

.pull-left[
This produces a reasonable plot:

.hide-code[

``` r
ggplot(co2new, aes(decimal_date, average)) +
    geom_line()
```

<img src="timeseries_files/figure-html/unnamed-chunk-18-1.png" style="display: block; margin: auto;" />
]
]
--

.pull-right[
The plot would show gaps for the missing values.
{{content}}
]
--

Missing values could also be interpolated with `na.interp` from
package `forecast`.

---

Converting the new data to a `ts` object makes adding the original
data a little easier:

``` r
co2newTS <- ts(co2new$average, start = c(1958, 3), frequency = 12)
autoplot(co2newTS) + autolayer(co2)
```

---
layout: true
## Seasonal Decomposition of Time Series by Loess (STL)
---

For time series with a strong seasonal component it can be useful to
look at a _Seasonal Decomposition of Time Series by Loess_, or
(_STL_).

The methodology was suggested by Clevaland and coworkers.

The `stl` function in the `base` package computes such a
decomposition.

It requires a series without missing values.

``` r
co2newTS_no_NA <- ts(co2new$average, start = c(1958, 3), frequency = 12)
co2stl <- stl(co2newTS_no_NA, s.window = 21)
head(co2stl$time.series)
## seasonal trend remainder
## Mar 1958 1.1159035 314.9426 -0.3484682
## Apr 1958 2.2797196 315.0143 0.1559367
## May 1958 2.7672696 315.0861 -0.3433923
## Jun 1958 2.2272621 315.1579 -0.1151637
## Jul 1958 0.8611739 315.2241 -0.2153025
## Aug 1958 -1.0848309 315.2904 0.7244755
```

---

.pull-left[
Package `forecast` provides an `autoplot` method for `stl` objects that
shows the original data and the three components in separate panels:

``` r
autoplot(co2stl)
```
]

.pull-right[
<img src="timeseries_files/figure-html/co2-stl-plot-1.png" style="display: block; margin: auto;" />
]

---
layout: false
## Some Alternative Visualizations

Bar charts are sometimes used when comparison to a base line is
important; for example for [illustrating climate
change](../img/CaG_GlobalTempAnom_1.jpg).

Calendar plots, or calendar heatmaps, are sometimes useful for
displaying daily data where day of the week/weekend effects might be
important.  One example is available [here](../img/calendarHeatmap.png).

A [blog post](https://dominikkoch.github.io/Calendar-Heatmap/) shows
how to build a calendar plot with `ggplot`.

---
layout: true
## GDP Growth Data
---

Bar charts are sometimes used for economic time series, in particular
growth rate data.

Some GDP growth rate data is available on this
[web page](https://www.multpl.com/us-real-gdp-growth-rate/table/by-quarter):

Read the data from the web page:

``` r
library(rvest)
gdpurl <- "http://www.multpl.com/us-real-gdp-growth-rate/table/by-quarter"
gdppage <- read_html(gdpurl)
gdptbl <- html_table(gdppage)[[1]]
head(gdptbl)
## # A tibble: 6 × 2
## Date Value
## <chr> <chr>
## 1 Dec 31, 2024 2.53%
## 2 Sep 30, 2024 2.72%
## 3 Jun 30, 2024 3.04%
## 4 Mar 31, 2024 2.90%
## 5 Dec 31, 2023 3.20%
## 6 Sep 30, 2023 3.24%
```
---

Clean the data:

``` r
library(dplyr)
library(lubridate)
names(gdptbl) <- c("Date", "Value")
gdptbl <- mutate(gdptbl,
 Value = as.numeric(sub("%", "", Value)),
 Date = mdy(Date))
head(gdptbl)
## # A tibble: 6 × 2
## Date Value
## <date> <dbl>
## 1 2024-12-31 2.53
## 2 2024-09-30 2.72
## 3 2024-06-30 3.04
## 4 2024-03-31 2.9 
## 5 2023-12-31 3.2 
## 6 2023-09-30 3.24
```

---

.pull-left[
Data prior to 1947 is annual, so drop earlier data:

``` r
gdptbl <- filter(gdptbl,
 Date >= make_date(1948, 1, 1))
```
{{content}}
]
--

Rearrange to put `Date` in increasing order:

``` r
gdptbl <- arrange(gdptbl, Date)
```
{{content}}
--

Create and plot a time series:

``` r
library(ggplot2)
library(forecast)
gdpts <- ts(gdptbl$Value,
 start = 1948, frequency = 4)
autoplot(gdpts) +
 geom_hline(yintercept = 0, linetype = 2)
```
--

.pull-right[
<img src="timeseries_files/figure-html/gdp-plot-1.png" style="display: block; margin: auto;" />
]

---

.pull-left[
A bar chart with color used to distinguish positive and negative
growth can be created from the original table `gdptbl`:

``` r
library(ggplot2)
mutate(gdptbl,
       growth = ifelse(Value >= 0,
                       "pos",
                       "neg")) |>
    ggplot(aes(x = Date,
               y = Value,
               fill = growth)) +
    geom_col() +
    theme(legend.position = "top") +
    scale_fill_manual(
        values = c(pos = "blue", neg = "red"))
```
]

.pull-right[
<img src="timeseries_files/figure-html/gdp-bar-1.png" style="display: block; margin: auto;" />
]

---

.pull-left[
The bar chart can also be created from the time series object:

``` r
data.frame(rate = as.numeric(gdpts),
           time = as.numeric(time(gdpts))) |>
    mutate(growth = ifelse(rate >= 0,
                           "pos",
                           "neg")) |>
    ggplot(aes(time, rate, fill = growth)) +
    geom_col() +
    theme(legend.position = "top") +
    scale_fill_manual(
        values = c(pos = "blue", neg = "red"))
```
]

.pull-right[
<img src="timeseries_files/figure-html/gdp-bar-ts-1.png" style="display: block; margin: auto;" />
]

---
layout: true
## Airline Flights From NYC
---

The `nycflights13` data set contains data on all flight leaving New
York City in 2013.

Similar data is available for other cities and years, e.g. from the
[`anyflights` package](https://simonpcouch.github.io/anyflights/).

We can look at the number of departures for each day:

``` r
library(nycflights13)
library(ggplot2)
nf <- count(flights, year, month, day)
```
--

It will be useful to have a combined `date` field in our data frame.

The `lubridate` function `make_date` can be used for this:

``` r
library(lubridate)
nf <- mutate(nf, date = make_date(year, month, day))
```

---

A line plot of the number of flights:

.pull-left[
<img src="timeseries_files/figure-html/flights-1-1.png" style="display: block; margin: auto;" />
]
--
.pull-right-width-40[

``` r
ggplot(nf, aes(date, n)) +
    geom_line()
```
]

.pull-right-width-40[
Weekday variation is clearly visible.
]

---

Breaking out the days of the week into separate facets may help:

.pull-left[
<img src="timeseries_files/figure-html/flights-2-1.png" style="display: block; margin: auto;" />
]

.pull-right[

``` r
nnf <-
* mutate(nf, wd = wday(date, label = TRUE))

nnf_thm <-
 theme(axis.text.x =
 element_text(angle = -45,
 hjust = 0),
 plot.margin = margin(r = 30))

ggplot(nnf, aes(date, n)) +
    geom_line() +
*   facet_wrap(~ wd) +
    nnf_thm
```
{{content}}
]
--

July 4 in 2013:

``` r
as.character(wday(ymd("2013-07-04"),
                  label = TRUE))
## [1] "Thu"
```
---

Identifying when some major holidays occur may also be useful:

.pull-left[
<img src="timeseries_files/figure-html/flights-3-1.png" style="display: block; margin: auto;" />
]
.pull-right[

``` r
holidays <-
 data.frame(Holiday = c("4th of July",
 "Thanksgiving"),
 date = ymd(c("2013-07-04",
 "2013-11-28")))
hlayer <-
 geom_vline(aes(xintercept = date,
 color = Holiday),
 linetype = 2,
 linewidth = 1,
 data = holidays)
ggplot(nf, aes(date, n)) +
 hlayer +
 geom_line()
```
]

---

.pull-left[
Holidays in the faceted plot:

<img src="timeseries_files/figure-html/flights-holiday-facet-1-1.png" style="display: block; margin: auto;" />
]

.pull-right[

``` r
ggplot(nnf, aes(date, n)) +
    hlayer +
    geom_line() +
    facet_wrap(~ wd) +
    nnf_thm
```
{{content}}
]
--

This marks the holiday weeks.

---

.pull-left[
Holidays in the faceted plot:

<img src="timeseries_files/figure-html/flights-holiday-facet-2-1.png" style="display: block; margin: auto;" />
--
]

.pull-right[
This shows only the holiday days.

``` r
hlayer2 <-
 geom_vline(aes(xintercept = date,
 color = Holiday),
 linewidth = 1,
 data = mutate(holidays,
 wd = wday(date,
 TRUE)))
ggplot(nnf, aes(date, n)) +
 hlayer2 +
 geom_line() +
 facet_wrap(~ wd) +
 nnf_thm
```
]

---

.pull-left[
Holidays in the faceted plot:

<img src="timeseries_files/figure-html/flights-holiday-facet-3-1.png" style="display: block; margin: auto;" />
]
.pull-right[
Show holidays as solid lines, days in holiday weeks as dashed lines.

``` r
ggplot(nnf, aes(date, n)) +
    hlayer +
    hlayer2 +
    geom_line() +
    facet_wrap(~ wd) +
    nnf_thm
```
]

---

A simple calendar plot:

.pull-left[
<img src="timeseries_files/figure-html/flights-cal-0-1.png" style="display: block; margin: auto;" />
]
--
.pull-right[
This uses `geom_tile` and faceting.

``` r
monthweek <- function(d, w)
 ceiling((d - w) / 7) + 1

nf <- mutate(nf,
 wd = wday(date, label = TRUE),
 wd = fct_rev(wd),
 mw = monthweek(day, wday(date)))

p <- ggplot(nf, aes(x = as.character(mw),
 y = wd,
 fill = n)) +
* geom_tile(color = "white") +
* facet_wrap(~ month(date, label = TRUE))
p
```
]

---

Some theme and color adjustments:

.pull-left[
<img src="timeseries_files/figure-html/flights-cal-1-1.png" style="display: block; margin: auto;" />
]
.pull-right[

``` r
p <- p +
 scale_fill_gradient(low = "white") +
 ylab("") + xlab("Week of Month") +
 theme(panel.grid.major = element_blank(),
 panel.border = element_blank())
p
```
]

---

This layout with days on the vertical axis works well with a single row,
which could be extended to a multi-year plot:

``` r
p + facet_wrap(~ month(date, TRUE), nrow = 1)
```

---

A more standard layout:

.pull-left[
<img src="timeseries_files/figure-html/flights-cal-3-1.png" style="display: block; margin: auto;" />
]
.pull-right[

``` r
nf2 <-
 mutate(nf,
 wd = wday(date, label = TRUE),
 mw = factor(monthweek(day,
 wday(date))),
 mw = fct_rev(mw))

wdlabs <-substr(levels(nf2$wd), 1, 1)

ggplot(nf2, aes(x = wd, y = mw, fill = n)) +
    geom_tile(color = "white") +
    facet_wrap(~ month(date, TRUE)) +
    scale_fill_gradient(low = "white") +
    xlab("") + ylab("Week of Month") +
    scale_x_discrete(labels = wdlabs) +
    theme(panel.grid.major = element_blank(),
          panel.border = element_blank())
```
]

---
layout: false
## Reading

Chapters [_Visualizing time series and other functions of an
  independent
  variable_](https://clauswilke.com/dataviz/time-series.html) and
  [_Visualizing
  trends_](https://clauswilke.com/dataviz/visualizing-trends.html) in
  [_Fundamentals of Data
  Visualization_](https://clauswilke.com/dataviz/).

//adapted from Emi Tanaka's gist at //https://gist.github.com/emitanaka/eaa258bb8471c041797ff377704c8505