San Francisco bike share network
Thuan Khang Dinh
2024-12-30
For Github version follow link
Introduction
This report analyzes the San Francisco bike share network using data from two datasets:
- July 7–13, 2014
- July 6–12, 2015
The goals are to evaluate the evolution of the network between 2014 and 2015 and identify critical stations or routes.
Data Loading
Loading the Data
stations = read.csv("https://raw.githubusercontent.com/thuankhang/casestudy-sf-bikeshare/refs/heads/main/data/SF-bikeshare-station-info.csv")
trips_2014 = read.csv("https://raw.githubusercontent.com/thuankhang/casestudy-sf-bikeshare/refs/heads/main/data/SF-bikeshare-1-week-2014-07.csv")
trips_2015 = read.csv("https://raw.githubusercontent.com/thuankhang/casestudy-sf-bikeshare/refs/heads/main/data/SF-bikeshare-1-week-2015-07.csv")Data Wrangling
Understand the dataset
We have 3 datasets:
stations:
str(stations)## 'data.frame': 70 obs. of 7 variables:
## $ id : int 2 3 4 5 6 7 8 9 10 11 ...
## $ name : chr "San Jose Diridon Caltrain Station" "San Jose Civic Center" "Santa Clara at Almaden" "Adobe on Almaden" ...
## $ lat : num 37.3 37.3 37.3 37.3 37.3 ...
## $ long : num -122 -122 -122 -122 -122 ...
## $ dock_count : int 27 15 11 19 15 15 15 15 15 19 ...
## $ city : chr "San Jose" "San Jose" "San Jose" "San Jose" ...
## $ installation_date: chr "8/6/2013" "8/5/2013" "8/6/2013" "8/5/2013" ...trips_2014:
str(trips_2014)## 'data.frame': 6911 obs. of 7 variables:
## $ start_date_yyyymmdd: chr "2014-07-13" "2014-07-13" "2014-07-13" "2014-07-13" ...
## $ start_station_name : chr "Powell at Post (Union Square)" "Market at 4th" "Market at 4th" "Grant Avenue at Columbus Avenue" ...
## $ start_station_id : int 71 76 76 73 50 2 61 75 28 71 ...
## $ end_date_yyyymmdd : chr "2014-07-13" "2014-07-13" "2014-07-13" "2014-07-13" ...
## $ end_station_name : chr "Embarcadero at Bryant" "Market at 10th" "Market at 10th" "Powell at Post (Union Square)" ...
## $ end_station_id : int 54 67 67 71 63 4 54 57 32 39 ...
## $ duration : int 667 401 401 470 421 221 233 455 559 1386 ...trips_2015:
str(trips_2015)## 'data.frame': 7381 obs. of 7 variables:
## $ start_date_yyyymmdd: chr "2015-07-12" "2015-07-12" "2015-07-12" "2015-07-12" ...
## $ start_station_name : chr "Howard at 2nd" "Temporary Transbay Terminal (Howard at Beale)" "San Jose City Hall" "Clay at Battery" ...
## $ start_station_id : int 63 55 10 41 77 42 16 16 76 50 ...
## $ end_date_yyyymmdd : chr "2015-07-12" "2015-07-12" "2015-07-12" "2015-07-12" ...
## $ end_station_name : chr "Market at Sansome" "Powell Street BART" "SJSU - San Salvador at 9th" "Washington at Kearny" ...
## $ end_station_id : int 77 39 16 46 73 49 10 16 69 56 ...
## $ duration : int 121 444 444 166 624 363 570 82 402 5868 ...Summary:
Let’s get the summary of trips_2014 and
trips_2015:
summary(trips_2014)## start_date_yyyymmdd start_station_name start_station_id end_date_yyyymmdd
## Length:6911 Length:6911 Min. : 2.00 Length:6911
## Class :character Class :character 1st Qu.:50.00 Class :character
## Mode :character Mode :character Median :62.00 Mode :character
## Mean :57.63
## 3rd Qu.:70.00
## Max. :84.00
## end_station_name end_station_id duration
## Length:6911 Min. : 2.00 Min. : 61
## Class :character 1st Qu.:49.00 1st Qu.: 348
## Mode :character Median :61.00 Median : 520
## Mean :57.23 Mean : 1158
## 3rd Qu.:70.00 3rd Qu.: 753
## Max. :84.00 Max. :715339summary(trips_2015)## start_date_yyyymmdd start_station_name start_station_id end_date_yyyymmdd
## Length:7381 Length:7381 Min. : 2.00 Length:7381
## Class :character Class :character 1st Qu.:50.00 Class :character
## Mode :character Mode :character Median :61.00 Mode :character
## Mean :57.65
## 3rd Qu.:70.00
## Max. :84.00
## end_station_name end_station_id duration
## Length:7381 Min. : 2.00 Min. : 61
## Class :character 1st Qu.:50.00 1st Qu.: 356
## Mode :character Median :61.00 Median : 531
## Mean :57.69 Mean : 1198
## 3rd Qu.:70.00 3rd Qu.: 768
## Max. :84.00 Max. :1133540NA observations
sum(is.na(stations))## [1] 0sum(is.na(trips_2014))## [1] 0sum(is.na(trips_2015))## [1] 0There is no missing(NA) values in any of the datasets.
Preprocessing
Standardize the datasets for consistent analysis.
Data Conversion and Cleanup
Add columns for start city, end city, and trip between cities.
trips_2014$start_city = stations$city[match(trips_2014$start_station_id, stations$id)]
trips_2014$end_city = stations$city[match(trips_2014$end_station_id, stations$id)]
trips_2014$trip_between_cities = paste(trips_2014$start_city, "->", trips_2014$end_city)
trips_2015$start_city = stations$city[match(trips_2015$start_station_id, stations$id)]
trips_2015$end_city = stations$city[match(trips_2015$end_station_id, stations$id)]
trips_2015$trip_between_cities = paste(trips_2015$start_city, "->", trips_2015$end_city)Add latitude and longitude information for each station.
trips_2014$start_station_lat = stations$lat[match(trips_2014$start_station_id, stations$id)]
trips_2014$start_station_long = stations$long[match(trips_2014$start_station_id, stations$id)]
trips_2014$end_station_lat = stations$lat[match(trips_2014$end_station_id, stations$id)]
trips_2014$end_station_long = stations$long[match(trips_2014$end_station_id, stations$id)]
trips_2015$start_station_lat = stations$lat[match(trips_2015$start_station_id, stations$id)]
trips_2015$start_station_long = stations$long[match(trips_2015$start_station_id, stations$id)]
trips_2015$end_station_lat = stations$lat[match(trips_2015$end_station_id, stations$id)]
trips_2015$end_station_long = stations$long[match(trips_2015$end_station_id, stations$id)]Analyze
a. Station Statistics
Overall:
We will find number of unique stations in July each year.
stations_2014 = unique(c(trips_2014$start_station_id, trips_2014$end_station_id))
stations_2015 = unique(c(trips_2015$start_station_id, trips_2015$end_station_id))- Number of stations in July 2014: 69.
- Number of stations in July 2015: 70.
There are more stations on July 2015, so we will find new stations and removed stations (if have) over the year:
new_stations = setdiff(stations_2015, stations_2014)
removed_stations = setdiff(stations_2014, stations_2015)- New stations in July 2015: Redwood City Medical Center
- Removed stations in July 2015:
By City:
First, we will need to find out how many cities are there, and what are those for later using.
city_2014 = unique(stations$city[stations$id %in% unique(union(trips_2014$start_station_id, trips_2014$end_station_id))])
city_2015 = unique(stations$city[stations$id %in% unique(union(trips_2015$start_station_id, trips_2015$end_station_id))])- In July 2014: Cities: San Jose, Redwood City, Mountain View, Palo Alto, San Francisco.
- In July 2015: Cities: San Jose, Redwood City, Mountain View, Palo Alto, San Francisco.
There are in total 5 cities in dataset of July 2014 and 5 cities in dataset of July 2015
- In July 2014:
stations_2014_count = table(stations$city[stations$id %in% stations_2014])
stations_2014_count##
## Mountain View Palo Alto Redwood City San Francisco San Jose
## 7 5 6 35 16write.csv(stations_2014_count, "num_stations_city_2014.csv")- In July 2015:
stations_2015_count = table(stations$city[stations$id %in% stations_2015])
stations_2015_count##
## Mountain View Palo Alto Redwood City San Francisco San Jose
## 7 5 7 35 16Insights
- As we found before, there are one more stations, and new station Redwood City Medical Center is in the Redwood City.
- Number of stations on each city still remains, except for Redwood City.
- San Francisco has the most number of stations, while Palo Alto has the least. We can make a prediction that the connectivity and volume in San Francisco will outweighted the others.Plotting
Pie chart number of stations each city over 2 years.
pie(stations_2014_count,
main = "Pie Chart of Stations in Cities - July 2014",
col = rainbow(length(stations_2014_count)),
radius = 1)
pie(stations_2015_count,
main = "Pie Chart of Stations in Cities - July 2015",
col = rainbow(length(stations_2015_count)),
radius = 1)
Grouped Bar Chart for comparing number of stations each city.
stations_count = data.frame(
City = rep(names(stations_2014_count), 2),
Year = rep(c(2014, 2015), each = length(stations_2014_count)),
Count = c(as.vector(stations_2014_count), as.vector(stations_2015_count))
)
ggplot(stations_count, aes(x = City, y = Count, fill = factor(Year))) +
geom_bar(stat = "identity", position = "dodge") +
theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
labs(x = "City", y = "Number of Stations", fill = "Year") +
ggtitle("Number of Stations by City in July 2014 and 2015") +
theme_minimal()
Graph for showing stations in map
Since the different between datasets in 2014 and 2015 is not huge enough, I will plot only graph for the year 2014.
First, get dataframe of station start, end and vol.
stations_graphing_2014 = as.data.frame(table(trips_2014$start_station_id, trips_2014$end_station_id))
colnames(stations_graphing_2014) = c("Start", "End", "Vol")Then, make graph using graph_from_data_frame
graph_2014 = graph_from_data_frame(stations_graphing_2014)
long_2014 = c()
lat_2014 = c()
for (station_id in V(graph_2014)$name) {
station_coords = stations[stations$id == station_id,]
if (nrow(station_coords) > 0) {
long_2014 = c(long_2014, station_coords$long)
lat_2014 = c(lat_2014, station_coords$lat)
} else {
long_2014 = c(long_2014, NA)
lat_2014 = c(lat_2014, NA)
}
}
V(graph_2014)$x = long_2014
V(graph_2014)$y = lat_2014
V(graph_2014)$size = degree(graph_2014, mode = "out")*0.1And plotting:
USA <- geodata::gadm(country='USA', level=2, path = ".")
cities_map <- USA[USA$NAME_1 %in% "California",]
plot(cities_map,
xlim = c(-122.45, -121.875),
ylim = c(37.325, 37.825))
plot(graph_2014, add = TRUE, rescale = FALSE,
edge.arrow.size = 0.01, edge.arrow.width = 0.75)
It is too difficult to see, so my first plan is I will delve into each city. But I cannot plot other city (I don’t know why) so I will plot only San Francisco.
The code are the same so I will skip explaining them.
San Francisco
stations_graphing_2014 = as.data.frame(table(trips_2014$start_station_id[trips_2014$start_city == "San Francisco" & trips_2014$end_city == "San Francisco"], trips_2014$end_station_id[trips_2014$start_city == "San Francisco" & trips_2014$end_city == "San Francisco"]))
colnames(stations_graphing_2014) = c("Start", "End", "Vol")
graph_2014 = graph_from_data_frame(stations_graphing_2014)
long_2014 = c()
lat_2014 = c()
for (station_id in V(graph_2014)$name) {
station_coords = stations[stations$id == station_id,]
if (nrow(station_coords) > 0) {
long_2014 = c(long_2014, station_coords$long)
lat_2014 = c(lat_2014, station_coords$lat)
} else {
long_2014 = c(long_2014, NA)
lat_2014 = c(lat_2014, NA)
}
}
V(graph_2014)$x = long_2014
V(graph_2014)$y = lat_2014
V(graph_2014)$size = degree(graph_2014, mode = "out")*0.01
USA <- geodata::gadm(country='USA', level=2, path = ".")
map <- USA[USA$NAME_2 == "San Francisco",]
plot(map,
xlim = c(-122.425, -122.375),
ylim = c(37.77, 37.81),
main = "Graph in San Francisco")
plot(graph_2014, add = TRUE, rescale = FALSE,
edge.arrow.size = 0.01, edge.arrow.width = 0.75)
b. Trip Volumes
Overall:
Total trips:
- Total trips in July 2014: 6911
- Total trips in July 2015: 7381
Total trips duration:
- Total trips duration in July 2014: 8003277 seconds, which is around 2223.13 hours.
- Total trips duration in July 2015: 8846153 seconds, which is 2457.26 hours.
Average trips duration:
- Average trip duration in July 2014: 1158.05 seconds.
- Average trip duration in July 2015: 1198.5 seconds.
Insights
- Total Trips: Increased from 6,911 in 2014 to 7,381 in 2015, reflecting greater bike-share usage.
- Total Trip Durations: Grew from 2,223.13 hours in 2014 to 2,457.26 hours in 2015, suggesting longer or more frequent trips.
- Average Trip Duration: Increased slightly from 1,158.05 seconds in 2014 to 1,198.5 seconds in 2015, indicating a trend toward longer rides.By city:
In July 2014:
table_trips_2014 = as.data.frame(table(trips_2014$trip_between_cities))
colnames(table_trips_2014) = c("trip_between_cities_2014", "num_trips")
duration_2014 = c()
for (i in table_trips_2014$trip_between_cities_2014) {
duration_2014 = c(duration_2014, sum(trips_2014$duration[trips_2014$trip_between_cities %in% i]))
}
table_trips_2014$duration = duration_2014
table_trips_2014$avg_duration = round(table_trips_2014$duration / table_trips_2014$num_trips, 2)
table_trips_2014## trip_between_cities_2014 num_trips duration avg_duration
## 1 Mountain View -> Mountain View 260 518743 1995.17
## 2 Mountain View -> Palo Alto 6 11596 1932.67
## 3 Palo Alto -> Mountain View 6 8959 1493.17
## 4 Palo Alto -> Palo Alto 86 144850 1684.30
## 5 Redwood City -> Palo Alto 2 6829 3414.50
## 6 Redwood City -> Redwood City 27 15796 585.04
## 7 San Francisco -> San Francisco 6071 5867541 966.49
## 8 San Jose -> San Jose 453 1428963 3154.44table_trips_2014$year = rep(2014, dim(table_trips_2014)[1])
table_trips_2014## trip_between_cities_2014 num_trips duration avg_duration year
## 1 Mountain View -> Mountain View 260 518743 1995.17 2014
## 2 Mountain View -> Palo Alto 6 11596 1932.67 2014
## 3 Palo Alto -> Mountain View 6 8959 1493.17 2014
## 4 Palo Alto -> Palo Alto 86 144850 1684.30 2014
## 5 Redwood City -> Palo Alto 2 6829 3414.50 2014
## 6 Redwood City -> Redwood City 27 15796 585.04 2014
## 7 San Francisco -> San Francisco 6071 5867541 966.49 2014
## 8 San Jose -> San Jose 453 1428963 3154.44 2014In July 2015:
table_trips_2015 = as.data.frame(table(trips_2015$trip_between_cities))
colnames(table_trips_2015) = c("trip_between_cities_2015", "num_trips")
duration_2015 = c()
for (i in table_trips_2015$trip_between_cities_2015) {
duration_2015 = c(duration_2015, sum(trips_2015$duration[trips_2015$trip_between_cities %in% i]))
}
table_trips_2015$duration = duration_2015
table_trips_2015$avg_duration = round(table_trips_2015$duration / table_trips_2015$num_trips, 2)
table_trips_2015## trip_between_cities_2015 num_trips duration avg_duration
## 1 Mountain View -> Mountain View 230 329696 1433.46
## 2 Mountain View -> Palo Alto 1 1917 1917.00
## 3 Palo Alto -> Mountain View 7 52995 7570.71
## 4 Palo Alto -> Palo Alto 81 1282512 15833.48
## 5 Redwood City -> Redwood City 47 139199 2961.68
## 6 San Francisco -> San Francisco 6628 6684147 1008.47
## 7 San Jose -> San Jose 387 355687 919.09table_trips_2015$year = rep(2015, dim(table_trips_2015)[1])
table_trips = data.frame(trip_between_cities = c(table_trips_2014$trip_between_cities_2014, table_trips_2015$trip_between_cities_2015),
num_trips = c(table_trips_2014$num_trips, table_trips_2015$num_trips),
duration = c(table_trips_2014$duration, table_trips_2015$duration),
avg_duration = c(table_trips_2014$avg_duration, table_trips_2015$avg_duration),
year = c(table_trips_2014$year, table_trips_2015$year))
write.csv(table_trips, "trips_summary_between_city_both.csv")Insights
1. 2014
- San Francisco had the highest number of trips (6,071) and significant total duration (5,867,541 seconds, avg. 966.49 seconds per trip). It was the busiest and most central hub.
- San Jose had notable trips (453) with a high average duration (3,154.44 seconds), reflecting longer commutes.
- Outliers like Redwood City -> Palo Alto had very high average durations (3,414.50 seconds) despite fewer trips (2 trips), likely due to recreational or long-distance rides.
3. 2015
- San Francisco again dominated with 6,628 trips, an increase from 2014, and a similar average trip duration (1,008.47 seconds).
- Palo Alto -> Palo Alto stood out with 81 trips and the highest average duration (15,833.48 seconds), suggesting unique, long-duration usage patterns.
- San Jose showed shorter average trips (919.09 seconds) despite a considerable number of trips (387), indicating a shift toward shorter commutes.PLotting:
Grouped Bar Chart for total number of trips and trip duration
data = data.frame(
Year = c("2014", "2015"),
Total_Trips = c(nrow(trips_2014), nrow(trips_2015)),
Total_Duration = c(sum(trips_2014$duration), sum(trips_2015$duration))
)
ggplot(data) +
geom_bar(aes(x = Year, y = Total_Trips, fill = "Total Trips"),
stat = "identity", position = "dodge", width = 0.4) +
geom_line(aes(x = Year, y = Total_Duration / 1000, group = 1, color = "Total Duration"),
linewidth = 1.5, linetype = "dashed") +
geom_point(aes(x = Year, y = Total_Duration / 1000, color = "Total Duration"),
size = 4) +
labs(title = "Total Trips and Total Duration in July 2014 and 2015",
x = "Year", y = "Value",
fill = "Legend", color = "Legend") +
scale_y_continuous(
name = "Total Trips",
sec.axis = sec_axis(~ . * 1000, name = "Total Duration (seconds)",
labels = scales::comma)
) +
scale_fill_manual(values = c("Total Trips" = "skyblue")) +
scale_color_manual(values = c("Total Duration" = "red")) +
theme_minimal()
Bar Chart for Average trip duration:
avg_data = data.frame(
Year = c("2014", "2015"),
Average_Duration = c(mean(trips_2014$duration), mean(trips_2015$duration))
)
ggplot(avg_data, aes(x = Year, y = Average_Duration, fill = Year)) +
geom_bar(stat = "identity") +
labs(title = "Average Trip Duration in July 2014 and 2015",
x = "Year", y = "Average Duration (seconds)") +
theme_minimal() +
scale_y_continuous(labels = scales::comma)
c. Routes Statistics
We will first make a data frame contains start_station, end_station, num_trips, duration and avg_duration for both year:
popular_routes_num_trips_2014 = aggregate(
duration ~ start_station_name + end_station_name,
data = trips_2014,
FUN = length
)
colnames(popular_routes_num_trips_2014)[3] = "num_trips_2014"
popular_routes_duration_2014 = aggregate(
duration ~ start_station_name + end_station_name,
data = trips_2014,
FUN = sum
)
colnames(popular_routes_duration_2014)[3] = "duration_2014"
popular_routes_2014 = merge(
popular_routes_num_trips_2014,
popular_routes_duration_2014,
by = c("start_station_name", "end_station_name")
)
popular_routes_2014$avg_duration_2014 = round(popular_routes_2014$duration_2014 / popular_routes_2014$num_trips_2014, 2)
popular_routes_num_trips_2015 = aggregate(
duration ~ start_station_name + end_station_name,
data = trips_2015,
FUN = length
)
colnames(popular_routes_num_trips_2015)[3] = "num_trips_2015"
popular_routes_duration_2015 = aggregate(
duration ~ start_station_name + end_station_name,
data = trips_2015,
FUN = sum
)
colnames(popular_routes_duration_2015)[3] = "duration_2015"
popular_routes_2015 = merge(
popular_routes_num_trips_2015,
popular_routes_duration_2015,
by = c("start_station_name", "end_station_name")
)
popular_routes_2015$avg_duration_2015 = round(popular_routes_2015$duration_2015 / popular_routes_2015$num_trips_2015, 2)Top 5 popular routes (by number of trips):
- In July 2014:
head(popular_routes_2014[order(-popular_routes_2014$num_trips_2014), ], 5)## start_station_name end_station_name
## 472 Harry Bridges Plaza (Ferry Building) Embarcadero at Sansome
## 383 Embarcadero at Sansome Steuart at Market
## 819 San Francisco Caltrain (Townsend at 4th) Embarcadero at Folsom
## 43 2nd at South Park Market at Sansome
## 595 Market at Sansome 2nd at South Park
## num_trips_2014 duration_2014 avg_duration_2014
## 472 82 107834 1315.05
## 383 50 20609 412.18
## 819 48 39512 823.17
## 43 46 18234 396.39
## 595 46 17118 372.13- In July 2015:
head(popular_routes_2015[order(-popular_routes_2015$num_trips_2015), ], 5)## start_station_name
## 466 Harry Bridges Plaza (Ferry Building)
## 1109 Townsend at 7th
## 68 2nd at Townsend
## 376 Embarcadero at Sansome
## 458 Harry Bridges Plaza (Ferry Building)
## end_station_name num_trips_2015 duration_2015
## 466 Embarcadero at Sansome 86 302489
## 1109 San Francisco Caltrain 2 (330 Townsend) 79 72478
## 68 Harry Bridges Plaza (Ferry Building) 78 61099
## 376 Steuart at Market 70 28426
## 458 2nd at Townsend 58 38755
## avg_duration_2015
## 466 3517.31
## 1109 917.44
## 68 783.32
## 376 406.09
## 458 668.19Top 5 routes with longest duration:
- In July 2014:
head(popular_routes_2014[order(-popular_routes_2014$duration_2014), ],5)## start_station_name end_station_name
## 114 Arena Green / SAP Center Adobe on Almaden
## 796 San Antonio Caltrain Station Castro Street and El Camino Real
## 784 Rengstorff Avenue / California Street Rengstorff Avenue / California Street
## 476 Harry Bridges Plaza (Ferry Building) Harry Bridges Plaza (Ferry Building)
## 892 San Jose City Hall Ryland Park
## num_trips_2014 duration_2014 avg_duration_2014
## 114 1 715339 715339.00
## 796 2 145090 72545.00
## 784 2 119454 59727.00
## 476 16 113310 7081.88
## 892 2 111430 55715.00- In July 2015:
head(popular_routes_2015[order(-popular_routes_2015$duration_2015), ], 5)## start_station_name end_station_name
## 1121 University and Emerson University and Emerson
## 466 Harry Bridges Plaza (Ferry Building) Embarcadero at Sansome
## 670 Mountain View City Hall Mountain View Caltrain Station
## 578 Market at 4th Market at 4th
## 391 Embarcadero at Vallejo Market at 4th
## num_trips_2015 duration_2015 avg_duration_2015
## 1121 10 1205549 120554.90
## 466 86 302489 3517.31
## 670 34 167330 4921.47
## 578 14 145523 10394.50
## 391 5 132222 26444.40Top 5 routes with longest average duration:
- In July 2014:
head(popular_routes_2014[order(-popular_routes_2014$avg_duration_2014), ],5)## start_station_name end_station_name
## 114 Arena Green / SAP Center Adobe on Almaden
## 796 San Antonio Caltrain Station Castro Street and El Camino Real
## 784 Rengstorff Avenue / California Street Rengstorff Avenue / California Street
## 892 San Jose City Hall Ryland Park
## 118 Arena Green / SAP Center San Jose Diridon Caltrain Station
## num_trips_2014 duration_2014 avg_duration_2014
## 114 1 715339 715339
## 796 2 145090 72545
## 784 2 119454 59727
## 892 2 111430 55715
## 118 2 108624 54312- In July 2015:
head(popular_routes_2015[order(-popular_routes_2015$avg_duration_2015), ], 5)## start_station_name end_station_name
## 1121 University and Emerson University and Emerson
## 127 Arena Green / SAP Center San Salvador at 1st
## 784 Redwood City Caltrain Station Redwood City Caltrain Station
## 391 Embarcadero at Vallejo Market at 4th
## 1119 University and Emerson Mountain View Caltrain Station
## num_trips_2015 duration_2015 avg_duration_2015
## 1121 10 1205549 120554.9
## 127 1 95920 95920.0
## 784 1 31381 31381.0
## 391 5 132222 26444.4
## 1119 2 46119 23059.5Insights
1. Most Popular Routes
- Harry Bridges Plaza (Ferry Building) to Embarcadero at Sansome was the most frequent route in both years (82 trips in 2014, 86 in 2015), solidifying its role as a central commuter hub.
- 2015 saw new popular routes like Townsend at 7th to San Francisco Caltrain 2, reflecting changing commuter patterns.
2. Routes with Longest Total Durations
- 2014: Outlier trips like Arena Green / SAP Center to Adobe on Almaden (715,339 seconds) dominated.
- 2015: University and Emerson to itself emerged as the longest total duration route (1,205,549 seconds), indicating increased recreational use.
3. Routes with Longest Average Durations
- Both years featured outliers (e.g., University and Emerson, Arena Green / SAP Center) with unusually long durations, likely non-commuter trips.Plotting
Bar Chart for Most Popular Routes
First, I will combine data for both years for comparison
popular_routes_combined = merge(
popular_routes_2014[, c("start_station_name", "end_station_name", "num_trips_2014")],
popular_routes_2015[, c("start_station_name", "end_station_name", "num_trips_2015")],
by = c("start_station_name", "end_station_name"),
all = TRUE
)
popular_routes_combined = popular_routes_combined[order(-popular_routes_combined$num_trips_2014, -popular_routes_combined$num_trips_2015), ]Then, I only take the top 5 popular routes:
popular_routes_top5 = head(popular_routes_combined, 5)
popular_routes_top5_long = data.frame(
route = paste(popular_routes_top5$start_station_name, "->", popular_routes_top5$end_station_name),
num_trips = c(popular_routes_top5$num_trips_2014, popular_routes_top5$num_trips_2015),
year = rep(c("2014", "2015"), each = 5)
)
popular_routes_top5_long$route = paste(
substr(popular_routes_top5$start_station_name, 1, 10),
"...",
"->",
substr(popular_routes_top5$end_station_name, 1, 10),
"..."
)And plotting
ggplot(popular_routes_top5_long, aes(x = route, y = num_trips, fill = year)) +
geom_bar(stat = "identity", position = "dodge") +
labs(
title = "Most Popular Routes by Number of Trips (Top 5)",
x = "Route",
y = "Number of Trips",
fill = "Year"
) +
theme(axis.text.x = element_text(angle = 60, hjust = 1))
Heatmap of top 100 Station Usage
First, I combine data for both years
heatmap_data_2014 <- trips_2014[, c("start_station_name", "end_station_name")]
heatmap_data_2015 <- trips_2015[, c("start_station_name", "end_station_name")]
heatmap_data <- rbind(heatmap_data_2014, heatmap_data_2015)And create heatmap data frame of top 25
heatmap_df <- as.data.frame(table(heatmap_data$start_station_name, heatmap_data$end_station_name))
colnames(heatmap_df) <- c("Start_Station", "End_Station", "Frequency")
heatmap_df = head(heatmap_df[order(-heatmap_df$Frequency), ], 25)And plotting:
ggplot(heatmap_df, aes(x = Start_Station, y = End_Station, fill = Frequency)) +
geom_tile() +
scale_fill_gradient(low = "white", high = "blue") +
labs(
title = "Heatmap of Top 25 Station Usage",
x = "Start Station",
y = "End Station",
fill = "Frequency"
) +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
Insights
- Station Statistics: The number of stations slightly increased from 69 in 2014 to 70 in 2015, with the addition of the Redwood City Medical Center station. San Francisco remains the central hub with the highest number of stations, while Palo Alto continues to have the fewest.
- Trip Volumes: Overall, trip volume and total duration grew from 6,911 trips (2,223.13 hours) in 2014 to 7,381 trips (2,457.26 hours) in 2015, reflecting greater bike-share usage. Average trip duration slightly increased, indicating longer rides over time.
- Routes Statistics: The most popular route was consistently between Harry Bridges Plaza and Embarcadero at Sansome, which remained the most traveled in both years. Routes with the longest durations were dominated by outliers, such as the Arena Green to Adobe on Almaden route in 2014 and University and Emerson in 2015, suggesting increased recreational or non-commuter use.
Overall, the data reflects growing bike-share usage, particularly in San Francisco, and a trend toward longer and more varied trips.