Code associated with blog post: http://t-redactyl.github.io/blog/2015/10/avoiding-the-crowds-in-melbourne-cbd-using-data.html

av_peds_2.R

av_peds_2 <- ddply(p.subset, c("date", "collapsed_sensors_2"), summarise,
                 n_peds = sum(Hourly_Counts))

# Extract weekday versus weekend
av_peds_2$day <- weekdays(av_peds_2$date, abbreviate = FALSE)
av_peds_2$weekend <- ifelse((av_peds_2$day == "Saturday" | av_peds_2$day == "Sunday"), 
                           "Weekend", "Weekday")
av_peds_2$weekend <- as.factor(av_peds_2$weekend)

# Extract time of day
av_peds_2$time=format(as.POSIXct(av_peds_2$date, format="%Y-%m-%d %H:%M"), format="%H:%M")

# Collapse data to mean number of pedestrians per site, hour and weekend/weekday
av_peds_2 <- ddply(av_peds_2, c("collapsed_sensors_2", "time", "weekend"), summarise,
                 mean_peds = mean(n_peds),
                 n_measurements = length(n_peds),
                 sum_peds = sum(n_peds)
)

average_peds_1.R

# Sum the collapsed categories
library(plyr)
av_peds <- ddply(p.subset, c("date", "collapsed_sensors"), summarise,
                 n_peds = sum(Hourly_Counts))

# Extract weekday versus weekend
av_peds$day <- weekdays(av_peds$date, abbreviate = FALSE)
av_peds$weekend <- ifelse((av_peds$day == "Saturday" | av_peds$day == "Sunday"), 
                           "Weekend", "Weekday")
av_peds$weekend <- as.factor(av_peds$weekend)

# Extract time of day
av_peds$time=format(as.POSIXct(av_peds$date, format="%Y-%m-%d %H:%M"), format="%H:%M")

# Collapse data to mean number of pedestrians per site, hour and weekend/weekday
av_peds <- ddply(av_peds, c("collapsed_sensors", "time", "weekend"), summarise,
                 mean_peds = mean(n_peds),
                 n_measurements = length(n_peds),
                 sum_peds = sum(n_peds))

cbd_centre_graph.R

city.centre <- av_peds_2[av_peds_2$collapsed_sensors_2 %in% 
                             c("Bourke Street Mall", "Chinatown-Swanston St (North)", 
                               "Swanston Street South", "Melbourne Central",
                               "Flinders Street-Elizabeth/Swanston Sts",
                               "QV Market-Elizabeth St (West)"), ]

ggplot(data=city.centre, aes(x=time, y=mean_peds, group=collapsed_sensors_2, colour=collapsed_sensors_2)) +
    geom_line() +
    geom_point() +
    facet_grid(. ~ weekend) +
    ylab("Mean pedestrians per hour") +
    xlab("Time of day") +
    scale_x_discrete(breaks=c("00:00", "03:00", "06:00", "09:00", 
                              "12:00", "15:00", "18:00", "21:00")) +
    theme_bw() +
    scale_shape_discrete(name = "City centre") +
    scale_colour_discrete(name = "City centre")

cbd_east_graph.R

# Extract data set with just the city east sites
city.east <-  av_peds[av_peds$collapsed_sensors %in% 
                          c("Alfred Place", "Lonsdale St-Spring St (West)", 
                            "Collins Place", "Flinders St-Spring Street",
                            "Bourke St-Russel St (West)", 
                            "Chinatown-Lt Bourke St (South)",
                            "Lonsdale Street (South)"),]
library(ggplot2)
ggplot(data=city.east, aes(x=time, y=mean_peds, 
                           group=collapsed_sensors, 
                           colour=collapsed_sensors)) +
    geom_line() +
    geom_point() +
    facet_grid(. ~ weekend) +
    ylab("Mean pedestrians per hour") +
    xlab("Time of day") +
    scale_x_discrete(breaks=c("00:00", "03:00", "06:00", "09:00", 
                              "12:00", "15:00", "18:00", "21:00")) +
    theme_bw() +
    scale_shape_discrete(name = "City East") +
    scale_colour_discrete(name = "City East")

centre_hour_average.R

centre.hour <- ddply(city.centre, c("time", "weekend"), summarise,
                     total_peds = sum(sum_peds),
                     total_measurements = sum(n_measurements)   
)
centre.hour$av_peds <- centre.hour$total_peds / centre.hour$total_measurements
max.weekday <- centre.hour$time[centre.hour$av_peds == 
                                    max(centre.hour$av_peds[
                                        centre.hour$weekend == "Weekday"])]
max.weekend <- centre.hour$time[centre.hour$av_peds == 
                                    max(centre.hour$av_peds[
                                        centre.hour$weekend == "Weekend"])]

east_hour_average.R

east.hour <- ddply(city.east, c("time", "weekend"), summarise,
                   total_peds = sum(sum_peds),
                   total_measurements = sum(n_measurements)   
)
east.hour$av_peds <- east.hour$total_peds / east.hour$total_measurements
max.weekday <- east.hour$time[east.hour$av_peds == 
                                  max(east.hour$av_peds[
                                      east.hour$weekend == "Weekday"])]
max.weekend <- east.hour$time[east.hour$av_peds == 
                                  max(east.hour$av_peds[
                                      east.hour$weekend == "Weekend"])]

getting_rid_public_holidays.R

p.subset <- with(p.subset, p.subset
                 [!(date >= "2013-11-05 00:00:00" & date <= "2013-11-05 23:00:00") & 
                  !(date >= "2013-12-25 00:00:00" & date <= "2013-12-26 23:00:00") &
                  !(date >= "2013-12-31 00:00:00" & date <= "2014-01-01 23:00:00") &
                  !(date >= "2014-01-27 00:00:00" & date <= "2014-01-27 23:00:00") &
                  !(date >= "2014-02-22 00:00:00" & date <= "2014-02-23 23:00:00") &
                  !(date >= "2014-03-10 00:00:00" & date <= "2014-03-10 23:00:00") &
                  !(date >= "2014-04-18 00:00:00" & date <= "2014-04-21 23:00:00") &
                  !(date >= "2014-04-25 00:00:00" & date <= "2014-04-25 23:00:00") &
                  !(date >= "2014-06-09 00:00:00" & date <= "2014-06-09 23:00:00") &
                  !(date >= "2014-09-26 00:00:00" & date <= "2014-09-26 23:00:00") &
                  !(date >= "2014-11-04 00:00:00" & date <= "2014-11-04 23:00:00") &
                  !(date >= "2014-12-25 00:00:00" & date <= "2014-12-26 23:00:00") &
                  !(date >= "2014-12-31 00:00:00" & date <= "2015-01-01 23:00:00") &
                  !(date >= "2015-01-26 00:00:00" & date <= "2015-01-26 23:00:00") &
                  !(date >= "2015-02-21 00:00:00" & date <= "2015-02-22 23:00:00") &
                  !(date >= "2015-03-09 00:00:00" & date <= "2015-03-09 23:00:00") &
                  !(date >= "2015-04-03 00:00:00" & date <= "2015-04-06 23:00:00") & 
                  !(date >= "2015-04-25 00:00:00" & date <= "2015-04-25 23:00:00") &
                  !(date >= "2015-06-08 00:00:00" & date <= "2015-06-08 23:00:00") &
                  !(date >= "2015-10-02 00:00:00" & date <= "2015-10-02 23:00:00"), ])

limiting_to_2_years.R

pedestrians$date <- as.POSIXct(as.character(pedestrians$Date_Time), 
                               format = "%d-%b-%Y %H:%M")
p.subset <- subset(pedestrians, date >= "2013-10-01 00:00:00")

loading_data.R

pedestrians <- read.csv(url("https://data.melbourne.vic.gov.au/api/views/b2ak-trbp/rows.csv"))

reducing_sensor_sites.R

p.subset <- p.subset[!p.subset$Sensor_Name %in% 
                         c("Australia on Collins", "Birrarung Marr", 
                         "Lygon St (East)", "Lygon Street (West)",
                         "Grattan St-Swanston St (West)",
                         "Melbourne Convention Exhibition Centre",
                         "Monash Rd-Swanston St (West)", "New Quay",
                         "Sandridge Bridge", "St Kilda-Alexandra Gardens",
                         "The Arts Centre", "Tin Alley-Swanston St (West)",
                         "Victoria Point", "Waterfront City", "Webb Bridge"), ]

p.subset$Sensor_Name <- as.character(p.subset$Sensor_Name)
p.subset$collapsed_sensors <- p.subset$Sensor_Name
p.subset$collapsed_sensors[
    p.subset$Sensor_Name == "Bourke Street Mall (North)" |
    p.subset$Sensor_Name == "Bourke Street Mall (South)"] <- "Bourke Street Mall"
p.subset$collapsed_sensors[
    p.subset$Sensor_Name == "City Square" |
    p.subset$Sensor_Name == "Town Hall (West)"] <- "Swanston Street South"
p.subset$collapsed_sensors[
    p.subset$Sensor_Name == "Collins Place (North)" |
    p.subset$Sensor_Name == "Collins Place (South)"] <- "Collins Place"
p.subset$collapsed_sensors[
    p.subset$Sensor_Name == "Flinders St-Spark La" |
    p.subset$Sensor_Name == "Flinders St-Spring St (West)"] <- "Flinders St-Spring St"
p.subset$collapsed_sensors[
    p.subset$Sensor_Name == "Flinders St-Elizabeth St (East)" |
    p.subset$Sensor_Name == "Flinders St-Swanston St (West)" |
    p.subset$Sensor_Name == "Flinders Street Station Underpass"] <- 
    "Flinders Street Station"
p.subset$collapsed_sensors[
    p.subset$Sensor_Name == "Southern Cross Station" |
    p.subset$Sensor_Name == "Spencer St-Collins St (North)" |
    p.subset$Sensor_Name == "Spencer St-Collins St (South)"] <- "Southern Cross Station"
p.subset$collapsed_sensors[
    p.subset$Sensor_Name == "Melbourne Central" |
    p.subset$Sensor_Name == "State Library"] <- "Melbourne Central"

p.subset$collapsed_sensors_2 <- p.subset$collapsed_sensors
p.subset$collapsed_sensors_2[
    p.subset$Sensor_Name == "Flinders St-Elizabeth St (East)" |
    p.subset$Sensor_Name == "Flinders St-Swanston St (West)"] <- 
    "Flinders Street-Elizabeth/Swanston Sts"
p.subset$collapsed_sensors_2[
    p.subset$Sensor_Name == "Flinders Street Station Underpass"] <- 
    "Flinders Street Station Underpass"

screening_data.R

head(pedestrians)
table(pedestrians$Sensor_Name)
tapply(pedestrians$Hourly_Counts, pedestrians$Sensor_Name, function(x) sum(is.na(x)))

stations_graph.R

# Extract data set with just the stations (except Parliament) and the surrounding sites
stations <- av_peds[av_peds$collapsed_sensors %in% 
                        c("Flinders Street Station", "Flagstaff Station", 
                          "Melbourne Central", "Southern Cross Station"), ]

# Graph of the average pedestrians at each station
library(ggplot2)
ggplot(data=stations, aes(x=time, y=mean_peds, 
                          group=collapsed_sensors, 
                          colour=collapsed_sensors)) +
    geom_line() +
    geom_point() +
    facet_grid(. ~ weekend) +
    ylab("Mean pedestrians per hour") +
    xlab("Time of day") +
    scale_x_discrete(breaks=c("00:00", "03:00", "06:00", "09:00", 
                              "12:00", "15:00", "18:00", "21:00")) +
    theme_bw() +
    scale_shape_discrete(name = "CBD Train Stations") +
    scale_colour_discrete(name = "CBD Train Stations")

stations_hour_average.R

station.hour <- ddply(stations, c("time", "weekend"), summarise,
                      total_peds = sum(sum_peds),
                      total_measurements = sum(n_measurements)   
)
station.hour$av_peds <- station.hour$total_peds / station.hour$total_measurements
max.weekday <- station.hour$time[station.hour$av_peds == 
                                     max(station.hour$av_peds[
                                         station.hour$weekend == "Weekday"])]
max.weekend <- station.hour$time[station.hour$av_peds == 
                                     max(station.hour$av_peds[
                                         station.hour$weekend == "Weekend"])]