1. Overview
The take-home exercise provides students the opportunity to revise and practice the R packages and programming skills we learnt in-class at home. This time,the exercise requires students to be innovative and creative by applying appropriate R packages to design enlightening and yet functional data visualization for analytics purposes. Students are encouraged to create multiple data visualization and compare our pros and cons before finalizing the best design.
2. Getting Started
Before we get started, it is important for us to ensure that the required R packages have been installed. If yes, we will load the R packages. If they have yet to be installed, we will install the R packages and load them onto R environment.
The chunk code below will do the trick.
packages = c('clock','lubridate', 'ggthemes', 'sftime',
'tidyverse', 'data.table', 'readr','sf','tmap','rmarkdown')
for (p in packages){
if(!require(p, character.only = T)){
install.packages(p)
}
library(p,character.only = T)
}
3. Importing Data
The code chunk below import TravelJournal.csv from the data
folder into R by using read_csv()
of readr
and save it as a tibble data frame called travel.
And use read_sf()
of sf,
saving it as data frame called pubs, schools,
apartment, buildings, employers,
restaurant
travel<- read_csv("data/TravelJournal.csv")
travel <- travel %>% mutate(travelEndLocationId = as.character(factor(travelEndLocationId))) %>%
select(-c(startingBalance,endingBalance,checkInTime,checkOutTime,travelStartLocationId,travelStartTime))
write_rds(travel,"data/travel.rds")
pubs <- read_sf("data/Pubs.csv", options = "GEOM_POSSIBLE_NAMES=location")
schools <- read_sf("data/Schools.csv", options = "GEOM_POSSIBLE_NAMES=location")
apartment <- read_sf("data/Apartments.csv", options = "GEOM_POSSIBLE_NAMES=location")
buildings <- read_sf("data/Buildings.csv", options = "GEOM_POSSIBLE_NAMES=location")
employers <- read_sf("data/Employers.csv", options = "GEOM_POSSIBLE_NAMES=location")
restaurant <- read_sf("data/Restaurants.csv", options = "GEOM_POSSIBLE_NAMES=location")
4.Data Visualizations and Insights
4.1 Social Area Detection
Generally the social area is physical or virtual space such as a social center, online social media, or other gathering place where people gather and interact. Thus, to identify physical social area in Ohio, we would get to know participants’ transport purpose. Since the TravelJournal already indicated one column with transport purpose, then we can leverage it to extract social purpose data for our specific target according.
To visualize the area, we still need to know polygon information, as TravelJournal itself doesn’t contain any polygon data, we need to join this journal with other table.
tmap_mode() is used to switch the display from static mode
tm_shape() is used to create a tmap-element that specifies a spatial data object
tm_polygon() is used to create a tmap-element that draws polygon feature.
write_csv(travel_social,"data/travel_social.csv")
social <- read_sf("data/travel_social.csv",
options = "GEOM_POSSIBLE_NAMES=location")
social <- social %>%
mutate(travelStartTime = date_time_parse(travelStartTime,zone = "",
format = "%Y-%m-%dT%H:%M:%S"))%>%
mutate(travelEndTime = date_time_parse(travelEndTime,zone = "",
format = "%Y-%m-%dT%H:%M:%S"))%>%
mutate(checkInTime = date_time_parse(checkInTime,zone = "",
format = "%Y-%m-%dT%H:%M:%S"))%>%
mutate(checkOutTime = date_time_parse(checkOutTime,zone = "",
format = "%Y-%m-%dT%H:%M:%S"))%>%
mutate(startDay = as.numeric(factor(startDay)))%>%
mutate(endDay = as.numeric(factor(endDay)))
social_select <- social %>%
select(-c(startingBalance,endingBalance,hourlyCost,maxOccupancy))
write_rds(social_select,"data/social_select.rds")
social_select<- read_rds("data/social_select.rds")
tm_shape(buildings)+
tm_polygons(col = "grey60",
size = 1.2,border.col = "black",border.lwd = 1) +
tm_shape(social_select) +
tm_dots(col = "red", size = 0.7, alpha = 0.2)
4.2 Traffic bottleneck
A traffic bottleneck is a localized disruption of vehicular traffic on a street, road, or highway. As opposed to a traffic jam, a bottleneck is a result of a specific physical condition. To define a reasonable metric of traffic bottleneck location, it seems reasonable to monitor number of commuter of a location per 10 minutes instead of total number of a whole day.
Another thing is that the traffic bottleneck would be caused by high volume of vehicle, which usually appears during peak times.Under this condition, we chop 7am-9am as morning peak time and 5pm-7pm as evening peak time. In addition, as the data sets don’t expose which type of vehicle does the commuter take, so assume part of them take public transport and others take private car. Then the destination with high volume would probably be traffic bottleneck.
As travel data frame doesn’t contain location information detail, we need to leverage and combine location from variable data frames.
apartment_ed <- apartment %>%
mutate(venueId = apartmentId) %>%
select(venueId,location)
employer_ed <- employers %>%
mutate(venueId = employerId) %>%
select(venueId,location)
pubs_ed<- pubs %>%
mutate(venueId = pubId) %>%
select(venueId,location)
school_ed<- schools %>%
mutate(venueId = schoolId) %>%
select(venueId,location)
restaurant_ed<- restaurant %>%
mutate(venueId = restaurantId) %>%
select(venueId,location)
combine <- rbind(apartment_ed,employer_ed,pubs_ed,school_ed,restaurant_ed)
Segregate moring peak and evening peak by time.
morning_peak <- read_rds("data/morning_peak.rds")
hex <- st_make_grid(buildings,cellsize=100,square=FALSE) %>%
st_sf() %>%
rowid_to_column('hex_id')
count_points_in_hex <- st_join(morning_peak,
hex,
join=st_within) %>%
st_set_geometry(NULL) %>%
count(name='pointCount', hex_id) %>%
mutate(count_per_10min = round(pointCount/(60*2)*10,2))
Morning peak traffic bottleneck location
During morning peak (7am-9am) high volume of vehicle appears in regional central area.
tmap_mode("plot")
tm_shape(hex_count %>%
filter(count_per_10min > 0))+
tm_fill("count_per_10min",
n = 5,
style = "quantile",
title = "Commuter/10min",palette="-RdBu") +
tm_borders(alpha = 0.3)+
tm_layout(title = "Morning peak bottleneck",title.size = 1, title.position = c("right","top"))
tmap_mode("plot")
Evening peak traffic bottleneck location
Comparing with morning peak, on contrast, during evening peak (5pm-7pm) high volume of vehicle appears in surrounding area of regional central. And from quantile of commuter per 10 minutes angle, evening peak’s highest value is greater than morning peak’s while the last quantile range of evening peak is broader, from 75 to 1376. Which means morning peak is more crowded than evening peak.
evening_peak <- read_rds("data/evening_peak.rds")
count_evening_in_hex <- st_join(evening_peak,
hex,
join=st_within) %>%
st_set_geometry(NULL) %>%
count(name='pointCount', hex_id) %>%
mutate(count_per_10min = round(pointCount/(60*2)*10,2))
tm_shape(hex_count_evening %>%
filter(count_per_10min > 0))+
tm_fill("count_per_10min",
n = 5,
style = "quantile",
title = "Commuter/10min",palette="-PiYG") +
tm_borders(alpha = 0.3)+
tm_layout(title = "Evening peak bottleneck",title.size = 1, title.position = c("right","top"))
