Calendar-based Graphics for Visualising People’s Daily Schedules

# Calendar-based Graphics for Visualising People’s Daily Schedules
### Earo Wang Monash University
### Dec 11, 2017

---

# .blue[Melbourne pedestrian activity]

---

.left-column[
## Pedestrian counting system
### - sensors
]
.right-column[
## The city of Melbourne
<img src="figure/sensor-map-1.svg" style="display: block; margin: auto;" />

]

---

.left-column[
## Pedestrian counting system
### - sensors
### - the data
]
.right-column[
* Sourced from [Melbourne open data portal](http://www.pedestrian.melbourne.vic.gov.au)
* Access the pedestrian data through the [rwalkr](http://pkg.earo.me/rwalkr/) package
* Up-to-date data till 2017-12-09

```
#> # A tibble: 378,024 x 6
#> Sensor Date_Time Date
#> <chr> <dttm> <date>
#> 1 Alfred Place 2017-01-01 2017-01-01
#> 2 Australia on Collins 2017-01-01 2017-01-01
#> 3 Birrarung Marr 2017-01-01 2017-01-01
#> 4 Bourke St-Russell St (West) 2017-01-01 2017-01-01
#> 5 Bourke Street Mall (North) 2017-01-01 2017-01-01
#> 6 Bourke Street Mall (South) 2017-01-01 2017-01-01
#> 7 Chinatown-Lt Bourke St (South) 2017-01-01 2017-01-01
#> 8 Chinatown-Swanston St (North) 2017-01-01 2017-01-01
#> 9 City Square 2017-01-01 2017-01-01
#> 10 Collins Place (North) 2017-01-01 2017-01-01
#> # ... with 378,014 more rows, and 3 more variables:
#> # Time <int>, Count <int>, Holiday <lgl>
```

]

---

.left-column[
## Take a glimpse
### - selected sensors
]
.right-column[
* Southbank
* Southern Cross Station
* Victoria Market
<img src="figure/selected-sensor-1.svg" style="display: block; margin: auto;" />
]

---

.left-column[
## Take a glimpse
### - selected sensors
### - time series plot
]
.right-column[
<img src="figure/ts-plot-1.svg" style="display: block; margin: auto;" />
]

---

.left-column[
## Take a glimpse
### - selected sensors
### - time series plot
### - faceted display
]
.right-column[
<img src="figure/facet-time-1.svg" style="display: block; margin: auto;" />
]

---

---

---

## calendar-based visualisation

---

.left-column[
## Calendar-based vis
### - construction
]
.right-column[
<center>
 <img src="img/month.png" width = 500>
</center>

The grid position for any day in the month is given by

`$$\begin{align}
i &= \lceil (g \mod 35) / 7\rceil \\ j &= g \mod 7.
\end{align}$$`

Let `$h$` and `$c$` be the scaled hour and count, respectively, then the final
coordinates are given by:

`$$\begin{align}
x &= j + h \\ y &= i - c.
\end{align}$$`
]

---

.left-column[
## Calendar-based vis
### - construction
### - the args
]
.right-column[
### The `frame_calendar()` function

```r
frame_calendar(
  data, x, y, date, calendar = "monthly", dir = "h",
  sunday = FALSE, nrow = NULL, ncol = NULL, polar = FALSE,
  scale = "fixed", width = 0.95, height = 0.95
)
```
* `x`, `y`: a unquoted (or bare) variable mapping to x and y axis.
* `date`: a Date variable mapping to dates in the calendar.
* `calendar`: type of calendar. "monthly", "weekly", "daily".
* `sunday`: `FALSE` indicating to starting with Monday in a week, or `TRUE` for Sunday.
* `nrow`, `ncol`: number of rows and columns defined for "monthly" calendar layout.
* `scale`: "fixed", "free", "free_wday", and "free_mday".
]

---

.left-column[
## Calendar-based vis
### - construction
### - the args
### - usage
]
.right-column[
## Rearrange the data
```r
sx_cal <- pedestrian %>% 
 filter(Sensor == "Southern Cross Station") %>% 
* frame_calendar(x = Time, y = Count, date = Date)
```

```
#> # A tibble: 8,232 x 5
#> Time Count Date .Time .Count
#> * <int> <int> <date> <dbl> <dbl>
#> 1 0 1335 2017-01-01 0.2583333 0.9860202
#> 2 1 463 2017-01-01 0.2598631 0.9721692
#> 3 2 219 2017-01-01 0.2613929 0.9682934
#> 4 3 122 2017-01-01 0.2629227 0.9667527
#> 5 4 21 2017-01-01 0.2644525 0.9651484
#> 6 5 28 2017-01-01 0.2659823 0.9652596
#> 7 6 26 2017-01-01 0.2675121 0.9652278
#> 8 7 35 2017-01-01 0.2690419 0.9653708
#> 9 8 55 2017-01-01 0.2705717 0.9656884
#> 10 9 82 2017-01-01 0.2721014 0.9661173
#> # ... with 8,222 more rows
```
]

---

.left-column[
## Calendar-based vis
### - construction
### - the args
### - usage
]
.right-column[
## Initialise the plot with `ggplot2`
```r
p1_sx <- sx_cal %>% 
 ggplot(aes(x = .Time, y = .Count, group = Date)) +
 geom_line()
```
<img src="figure/sx-plot-1.svg" style="display: block; margin: auto;" />
]

---

.left-column[
## Calendar-based vis
### - construction
### - the args
### - usage
]
.right-column[
## Add reference lines and labels
```r
prettify(p1_sx)
```
<img src="figure/sx-prettify-1.svg" style="display: block; margin: auto;" />
]

---

.left-column[
## Calendar-based vis
### - construction
### - the args
### - usage
]
.right-column[
## Colour mapping as usual 
```r
p2_sx <- sx_cal %>%
 ggplot(aes(x = .Time, y = .Count, colour = Holiday, group = Date)) +
 geom_line()
prettify(p2_sx)
```
<img src="figure/sx-hol-1.svg" style="display: block; margin: auto;" />
]

---

.left-column[
## Calendar-based vis
### - construction
### - the args
### - usage
]
.right-column[
## Zoom in to March
<img src="figure/sx-march-1.svg" style="display: block; margin: auto;" />
]

---

.left-column[
## Calendar-based vis
### - construction
### - the args
### - usage
]
.right-column[
## Naturally work with the `group_by()`
```r
facet_cal <- subdat %>% 
 group_by(Sensor) %>% 
* frame_calendar(
* x = Time, y = Count, date = Date, nrow = 2
* )

p_facet <- facet_cal %>% 
 ggplot(aes(x = .Time, y = .Count, group = Date)) +
 geom_line(aes(colour = Sensor)) +
 facet_grid(Sensor ~ .) +
 scale_colour_brewer(palette = "Dark2")
prettify(p_facet, label = NULL)
```
]

---

.left-column[
## Calendar-based vis
### - construction
### - the args
### - usage
]
.right-column[
## Faceted calendar plots
<img src="figure/facet-1.svg" style="display: block; margin: auto;" />
]

---

## Summary

* The full range of plotting capabilities in *ggplot2* is essentially available, from `geom_point()` to `geom_boxplot()`.
* Other language supports are made for month and weekday labels, like CJK.
* Patterns on special events for the region, like Anzac Day in Australia, or Thanksgiving Day in the USA, more easily pop out to the viewer as public holidays.
* This sort of layout will be useful for studying consumer trends, or human behavior, such as pedestrian patterns or residential electricity demand.

---

# Joint work with [Di Cook](http://dicook.org) and [Rob J Hyndman](http://robjhyndman.com)

# Made with []() and

# Questions