Take-home Exercise 5

Practical Visual Analytics Use: VAST Challenge 2022, Challenge 2 - Pattern of Life, focusing on map of social areas and locations with traffic bottleneck.

1. Overview

In this take-home exercise, we reveal the patterns of life in Ohio, USA by creating data visualization with tmap.

With reference to point 1 and 2 in Challenge 2 of VAST Challenge 2022, the following questions will be addressed:

• Assuming the volunteers are representative of the city’s population, characterize the social areas of the city that you identify.

• Where are the busiest areas in Engagement? Are there traffic bottlenecks that should be addressed?

2. Data Preparation

2.1 Installing Packages

The following code chunk installs the required R packages and loads them into RStudio environment.

packages = c('sf', 'tmap', 'tidyverse', 'lubridate', 'clock',
             'sftime', 'rmarkdown', 'dplyr')

for(p in packages){
  if(!require(p, character.only = T)){
    install.packages(p)
  }
  library(p, character.only = T)
}

2.2 Importing the Dataset

Relevant datasets are imported using read_sf() of sf package, which designs to handle, process, visualise and analyse movement or geospatial data.

buildings <- read_sf("data/Buildings.csv", 
                   options = "GEOM_POSSIBLE_NAMES=location")

employers <- read_sf("data/Employers.csv", 
                   options = "GEOM_POSSIBLE_NAMES=location")

pubs <- read_sf("data/Pubs.csv", 
                   options = "GEOM_POSSIBLE_NAMES=location")

restaurants <- read_sf("data/Restaurants.csv", 
                   options = "GEOM_POSSIBLE_NAMES=location")

apartments <- read_sf("data/Apartments.csv", 
                   options = "GEOM_POSSIBLE_NAMES=location")

2.3 Data Wrangling

To observe residents’ movement in Ohio everyday, we import an additional data, TravelJournal, for the following visualizations. The visualization of traffic bottleneck would divide into recreation and commute area. Followed by this, we would reveal the movement path of residents with the highest and lowest joviality to see the differences of their daily patterns.

2.3.1 Traffic Bottleneck on Recreation Areas

IMPORT TRAVEL JOURNAL

travel <- read_csv("data/TravelJournal.csv") %>%
  mutate(travelEndLocationId = as.character(travelEndLocationId),
         travelStartLocationId = as.character(travelStartLocationId))

FILTER RECREATION PURPOSE

travel1 <- travel %>%
  filter(purpose == "Recreation (Social Gathering)")

travel2 <- travel %>%
  filter(purpose == "Coming Back From Restaurant")

CREATE LOCATION POINTS DATA

restaurants_location <- restaurants %>%
  select(restaurantId, location, buildingId) %>%
  rename("locationId" = "restaurantId")

pubs_location <- pubs %>%
  select(pubId, location, buildingId) %>%
  rename("locationId" = "pubId")

apartments_location <- apartments %>%
  select(apartmentId, location, buildingId) %>%
  rename("locationId" = "apartmentId")

employers_location <- employers %>%
  select(employerId, location, buildingId) %>%
  rename("locationId" = "employerId")

social_area <- rbind(restaurants_location, 
                     pubs_location, 
                     apartments_location, 
                     employers_location)

JOIN LOCATION COLUMN

travel1 <- social_area %>%
  inner_join(travel1, by = c("locationId" = "travelEndLocationId")) %>%
  rename("travelEndLocationId" = "locationId")

travel2 <- social_area %>%
  inner_join(travel2, by = c("locationId" = "travelStartLocationId")) %>%
  rename("travelStartLocationId" = "locationId")

travel1 <- travel1[, c(4,5,6,7,1,8,9,10,11,12,2,3)]
travel2 <- travel2[, c(4,5,1,6,7,8,9,10,11,12,2,3)]

recreation <- rbind(travel1, travel2)

DROP USELESS COLUMNS

recreation <- recreation %>%
  select(-c(checkInTime, checkOutTime, startingBalance, endingBalance, 
            travelEndTime, buildingId, purpose))

DERIVE TIME FIELDS

recreation <- recreation %>%
  mutate(date = as.Date(travelStartTime)) %>%
  mutate(day = weekdays(date)) %>%
  mutate(days = get_day(travelStartTime)) 

FILTER WEEKDAY & WEEKEND

recreation_weekday <- recreation %>%
  filter(day == "Monday" | day == "Tuesday" | day == "Wednesday" |
           day == "Thursday" | day == "Friday")

recreation_weekend <- recreation %>%
  filter(day == "Saturday" | day == "Sunday")

2.3.2 Traffic Bottleneck on Commute Areas

FILTER COMMUTE PURPOSE

travel3 <- travel %>%
  filter(purpose == "Work/Home Commute")

JOIN LOCATION COLUMN

travel3 <- social_area %>%
  inner_join(travel3, by = c("locationId" = "travelEndLocationId")) %>%
  rename("travelEndLocationId" = "locationId")

travel3 <- travel3[, c(4,5,6,7,1,8,9,10,11,12,2,3)]

DROP USELESS COLUMNS

travel3 <- travel3 %>%
  select(-c(checkInTime, checkOutTime, startingBalance, endingBalance, 
            travelEndTime, buildingId, purpose))

DERIVE TIME FIELD

commute <- commute %>%
  mutate(date = as.Date(travelStartTime)) %>%
  mutate(day = weekdays(date)) %>%
  mutate(days = get_day(travelStartTime)) 

FILTER WEEKDAY & WEEKEND

commute_weekday <- commute %>%
  filter(day == "Monday" | day == "Tuesday" | day == "Wednesday" |
           day == "Thursday" | day == "Friday")

commute_weekend <- commute %>%
  filter(day == "Saturday" | day == "Sunday")

2.3.3 Movement Path

JOIN LOCATION COLUMN

movement <- social_area %>%
  inner_join(movement, by = c("locationId" = "travelEndLocationId")) %>%
  rename("travelEndLocationId" = "locationId")

movement <- movement[, c(4,5,6,7,1,8,9,10,11,12,2,3)]

DERIVE TIME FIELD

move_path <- movement %>%
  mutate(date = as.Date(travelStartTime)) %>%
  mutate(day = weekdays(date)) %>%
  mutate(days = get_day(travelStartTime)) 

FILTER PARTICIPANTS WITH HIGHEST AND LOWEST JOVIALITY

move_path <- move_path %>% 
  filter(participantId == "758" | participantId == "131") 

3. Data Visualizations and Insights

3.1 Social Areas

3.1.1 Social Areas in Residential Districts

• Blue dot: Residential Districts
• Red dot: Restaurants
• Yellow dot: Pubs

tmap_mode("plot")
tm_shape(buildings) +
tm_polygons(col = "grey60",
           size = 1,
           border.col = "black",
           border.lwd = 1) +
tm_shape(apartments) +
  tm_dots(col = "#4682B4", size = 0.3, alpha= 0.8) +
tm_shape(pubs) +
  tm_dots(col = "#FFD700", size = 0.5) +
tm_shape(restaurants) +
  tm_dots(col = "#DC143C", size = 0.5)

3.1.2 Social Areas in Workplaces

• Green dot: Workplaces
• Red dot: Restaurants
• Yellow dot: Pubs

tmap_mode("plot")
tm_shape(buildings)+
tm_polygons(col = "grey60",
           size = 1,
           border.col = "black",
           border.lwd = 1) +
tm_shape(employers) +
  tm_dots(col = "#2E8B57", size = 0.5) +
tm_shape(pubs) +
  tm_dots(col = "#FFD700", size = 0.5) +
tm_shape(restaurants) +
  tm_dots(col = "#DC143C", size = 0.5)

Insights

Compared to the distance between residential and recreation places, the location of workplaces has more overlapped areas. This is probably for office workers to release their pressure after work without moving a long distance to the recreational merchants.

3.2 Traffic Bottleneck on Recreation Areas

COMPUTE HEXAGONS

To plot hexagon binning maps, we need to compute and create hexagons first.

hex <- st_make_grid(buildings, 
                    cellsize=100, 
                    square=FALSE) %>%
  st_sf() %>%
  rowid_to_column('hex_id')

3.2.1 Weekday Recreation

COUNT EVENT POINTS

weekday_points <- st_join(recreation_weekday, 
                          hex, 
                          join = st_within) %>%
  st_set_geometry(NULL) %>%
  count(name ='pointCount', hex_id)

head(weekday_points)

# A tibble: 6 × 2
  hex_id pointCount
   <int>      <int>
1    641       6706
2    775      12388
3    815      28020
4    863      13217
5   1004      13824
6   1179       4834

PERFORM RELATIONAL JOIN

hex_weekday <- hex %>%
  left_join(weekday_points, 
            by = 'hex_id') %>%
  replace(is.na(.), 0)

VISUALISATION

tmap_mode("view")
tm_shape(hex_weekday %>%
           filter(pointCount > 0))+
  tm_fill("pointCount",
          n = 8,
          style = "quantile") +
  tm_borders(alpha = 0.1)

tmap_mode("plot")

3.2.1 Weekend Recreation

COUNT EVENT POINTS

weekend_points <- st_join(recreation_weekend, 
                          hex, 
                          join = st_within) %>%
  st_set_geometry(NULL) %>%
  count(name ='pointCount', hex_id)

head(weekend_points)

# A tibble: 6 × 2
  hex_id pointCount
   <int>      <int>
1    641       8613
2    775       3515
3    815      10684
4    863      13852
5   1004      14982
6   1179       1362

PERFORM RELATIONAL JOIN

hex_weekend <- hex %>%
  left_join(weekend_points, 
            by = 'hex_id') %>%
  replace(is.na(.), 0)

VISUALISATION

tmap_mode("view")
tm_shape(hex_weekend %>%
           filter(pointCount > 0))+
  tm_fill("pointCount",
          n = 8,
          style = "quantile") +
  tm_borders(alpha = 0.1)

tmap_mode("plot")

Insights

The popular places in recreation areas on weekend and weekday are different. In addition, it is also interesting that the event points on weekday are slightly more than point on weekend, meaning that people are more frequent to entertain in social areas after work.

3.2 Traffic Bottleneck on Commute Areas

3.2.1 Weekday Commute

COUNT EVENT POINTS

cweekday_points <- st_join(commute_weekday, 
                          hex, 
                          join = st_within) %>%
  st_set_geometry(NULL) %>%
  count(name ='pointCount', hex_id)

head(cweekday_points)

# A tibble: 6 × 2
  hex_id pointCount
   <int>      <int>
1    169        257
2    212        964
3    225        194
4    226       2119
5    227        642
6    228        964

PERFORM RELATIONAL JOIN

hex_cweekday <- hex %>%
  left_join(cweekday_points, 
            by = 'hex_id') %>%
  replace(is.na(.), 0)

VISUALISATION

tmap_mode("view")
tm_shape(hex_cweekday %>%
           filter(pointCount > 0))+
  tm_fill("pointCount",
          n = 8,
          style = "quantile") +
  tm_borders(alpha = 0.1)

tmap_mode("plot")

3.2.1 Weekdend Commute

COUNT EVENT POINTS

cweekend_points <- st_join(commute_weekend, 
                          hex, 
                          join = st_within) %>%
  st_set_geometry(NULL) %>%
  count(name ='pointCount', hex_id)

head(cweekend_points)

# A tibble: 6 × 2
  hex_id pointCount
   <int>      <int>
1    169         64
2    225        128
3    226        128
4    260        128
5    272        128
6    273         64

PERFORM RELATIONAL JOIN

hex_cweekend <- hex %>%
  left_join(cweekend_points, 
            by = 'hex_id') %>%
  replace(is.na(.), 0)

VISUALISATION

tmap_mode("view")
tm_shape(hex_cweekend %>%
           filter(pointCount > 0))+
  tm_fill("pointCount",
          n = 8,
          style = "quantile") +
  tm_borders(alpha = 0.1)

tmap_mode("plot")

Insights

From the above two graphs, it can obviously be observed that the issue of traffic congestion is more serious from Monday to Friday. Crowds of people head to those places, which leads to more transportation issues in the city and have to be addressed by the government.

3.3 Movement Path

CREATE MOVEMENT PATH FROM EVENT POINTS

path <- move_path %>%
  group_by(participantId, day) %>%
  summarize(m = mean(travelStartTime), 
            do_union=FALSE) %>%
  st_cast("LINESTRING")

3.3.1 Movement of Most Happiness Participants

• Blue dot: Residential Districts
• Green dot: Workplaces
• Red dot: Restaurants
• Yellow dot: Pubs

path_758 <- path %>%
  filter(participantId == 758)

tm_shape(buildings)+
tm_polygons(col = "grey60",
           size = 1,
           border.col = "black",
           border.lwd = 1) +
tm_shape(path_758) +
  tm_lines(col = '#ffa500',
           scale = 3) +
tm_shape(apartments) +
  tm_dots(col = "#4682B4", size = 0.3, alpha= 0.8) +
tm_shape(employers) +
  tm_dots(col = "#2E8B57", size = 0.5) +
tm_shape(pubs) +
  tm_dots(col = "#FFD700", size = 0.5) +
tm_shape(restaurants) +
  tm_dots(col = "#DC143C", size = 0.5)

3.3.1 Movement of Least Happiness Participants

• Blue dot: Residential Districts
• Green dot: Workplaces
• Red dot: Restaurants
• Yellow dot: Pubs

path_131 <- path %>%
  filter(participantId == 131)

tm_shape(buildings)+
tm_polygons(col = "grey60",
           size = 1,
           border.col = "black",
           border.lwd = 1) +
tm_shape(path_131) +
  tm_lines(col = '#ffa500',
           scale = 3) +
tm_shape(apartments) +
  tm_dots(col = "#4682B4", size = 0.3, alpha= 0.8) +
tm_shape(employers) +
  tm_dots(col = "#2E8B57", size = 0.5) +
tm_shape(pubs) +
  tm_dots(col = "#FFD700", size = 0.5) +
tm_shape(restaurants) +
  tm_dots(col = "#DC143C", size = 0.5)

Insights

There is no intersection of the most and least happiness participants in Ohio in the social cycle. Participant id = 758 with the highest joviality mainly has activities in the northwest of the city, while participant id = 131 with the lowest joviality activate in the east of south. In addition, participant id = 758 are more common to go to restaurants and pubs.

4. Learning Points

This take-home exercise helps us to understand how to plot visualizations for maps and geospatial data. Furthermore, it also makes me learn a new package in R for plotting a graph, `tmap``.

My key takeaways are:

• The types of spatial objects would influence the design of graphs. Thus, it is necessary for us to print and read the data to check the spatial objects before creating the visualization. In addition, the type of spatial objects can only be converted from bigger objects to smaller objects but cannot convert reversely. For example, it is feasible to convert objects from polygons to lines or points, but it not viable to obtain lines or polygons from points.

• It is always better to use the installed packages to clean data such as tidyverse and dplyr instead of using built-in function in R, since those functions are not always applicable to all data types in R. To achieve our data preparation process, we need to take more notice on picking the right function in our code.