Joris Gauliard 2022-09-01
Cyclistic is a fictional bike-share company based in Chicago. The company owns about 5800 bicycles and 600 docking stations.
In-house finance analysts have recently concluded that annual members are much more profitable than casual riders.As a result, the strategy of the director of marketing is clear: converting casual riders into annual members to increase the profitability of the company.
In order to better understand why casual riders would buy a membership, the marketing team is interested in analyzing the historical bike trip data to identify trends.
The result of this analysis, in the form of a data-driven marketing strategy proposal, will be submitted to the executive team for its approval.
The scope of this study is to understand how annual members and casual riders use Cyclistic bikes differently.
The historical bike trip data used in this study has been made available by Motivate International Inc. under this licence.
Data starting from January 2021 has been downloaded and stored in a folder.
Libraries are loaded for future use
Files are combined into one main dataframe (df) using reader::read_csv() and base::rbind()
filenames <- list.files(path="data", pattern="*.csv", full.names=TRUE)
df <- do.call("rbind", lapply(filenames, read_csv, show_col_types = FALSE))Summary functions help to get to know the dataset
dt.haversine <- function(lat_from, lon_from, lat_to, lon_to, r = 6378137){
radians <- pi/180
lat_to <- lat_to * radians
lat_from <- lat_from * radians
lon_to <- lon_to * radians
lon_from <- lon_from * radians
dLat <- (lat_to - lat_from)
dLon <- (lon_to - lon_from)
a <- (sin(dLat/2)^2) + (cos(lat_from) * cos(lat_to)) * (sin(dLon/2)^2)
return(2 * atan2(sqrt(a), sqrt(1 - a)) * r)
}
df$ride_distance <- dt.haversine(df$start_lat,df$start_lng,df$end_lat,df$end_lng)
df$ride_distance <- as.integer(df$ride_distance)#skim(df)
#head(df)
str(df)## spec_tbl_df [8,697,283 × 14] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
## $ ride_id : chr [1:8697283] "E19E6F1B8D4C42ED" "DC88F20C2C55F27F" "EC45C94683FE3F27" "4FA453A75AE377DB" ...
## $ rideable_type : chr [1:8697283] "electric_bike" "electric_bike" "electric_bike" "electric_bike" ...
## $ started_at : POSIXct[1:8697283], format: "2021-01-23 16:14:19" "2021-01-27 18:43:08" ...
## $ ended_at : POSIXct[1:8697283], format: "2021-01-23 16:24:44" "2021-01-27 18:47:12" ...
## $ start_station_name: chr [1:8697283] "California Ave & Cortez St" "California Ave & Cortez St" "California Ave & Cortez St" "California Ave & Cortez St" ...
## $ start_station_id : chr [1:8697283] "17660" "17660" "17660" "17660" ...
## $ end_station_name : chr [1:8697283] NA NA NA NA ...
## $ end_station_id : chr [1:8697283] NA NA NA NA ...
## $ start_lat : num [1:8697283] 41.9 41.9 41.9 41.9 41.9 ...
## $ start_lng : num [1:8697283] -87.7 -87.7 -87.7 -87.7 -87.7 ...
## $ end_lat : num [1:8697283] 41.9 41.9 41.9 41.9 41.9 ...
## $ end_lng : num [1:8697283] -87.7 -87.7 -87.7 -87.7 -87.7 ...
## $ member_casual : chr [1:8697283] "member" "member" "member" "member" ...
## $ ride_distance : int [1:8697283] 2244 556 280 2250 276 4541 1098 1103 267 2034 ...
## - attr(*, "spec")=
## .. cols(
## .. ride_id = col_character(),
## .. rideable_type = col_character(),
## .. started_at = col_datetime(format = ""),
## .. ended_at = col_datetime(format = ""),
## .. start_station_name = col_character(),
## .. start_station_id = col_character(),
## .. end_station_name = col_character(),
## .. end_station_id = col_character(),
## .. start_lat = col_double(),
## .. start_lng = col_double(),
## .. end_lat = col_double(),
## .. end_lng = col_double(),
## .. member_casual = col_character()
## .. )
## - attr(*, "problems")=<externalptr>
#glimpse(df)
#colnames(df)
#skim_without_charts(df)We start by removing the columns we do not need for this analysis. We then create few date-related columns to be able to identify ride length patterns as a function of those attributes. We sort the data by ascending ride legnth
Sys.setlocale("LC_TIME", "en_US.UTF-8")## [1] "en_US.UTF-8"
df_v2<-
df%>%
select(-c(start_lat,start_lng,end_lat,end_lng,start_station_name,start_station_id,end_station_name,end_station_id))%>%
mutate(ride_length = difftime(ended_at,started_at))%>%
mutate(date = as.Date(started_at))%>%
mutate(year = format(as.Date(started_at), "%Y"))%>%
mutate(month = format(as.Date(started_at), "%m"))%>%
mutate(day_of_week = weekdays(started_at))%>%
mutate(hour_of_day = hour(started_at))%>%
arrange(ride_length)
df_v2 <- df_v2[!(df_v2$ride_length <0 ),] #removing observations with a negative ride length
# Convert ride length from factor to numeric
is.factor(df_v2$ride_length)## [1] FALSE
df_v2$ride_length <- as.numeric(as.character(df_v2$ride_length))
is.numeric(df_v2$ride_length)## [1] TRUE
df_v2 <- df_v2%>%
select(-c(started_at,ended_at))
colnames(df_v2)## [1] "ride_id" "rideable_type" "member_casual" "ride_distance"
## [5] "ride_length" "date" "year" "month"
## [9] "day_of_week" "hour_of_day"
Checking if the dataframe contains NA
colSums(is.na(df_v2))## ride_id rideable_type member_casual ride_distance ride_length
## 0 0 0 8240 0
## date year month day_of_week hour_of_day
## 0 0 0 0 0
df_v2 <- na.omit(df_v2)
#colSums(is.na(df_v2))We can now consider that the data is prepared for further analysis. The dataset contains 8688865 observations.
head(df_v2)## # A tibble: 6 × 10
## ride_id ridea…¹ membe…² ride_…³ ride_…⁴ date year month day_o…⁵ hour_…⁶
## <chr> <chr> <chr> <int> <dbl> <date> <chr> <chr> <chr> <int>
## 1 6B5129… classi… member 259 0 2021-01-15 2021 01 Friday 16
## 2 3F0277… classi… member 0 0 2021-01-29 2021 01 Friday 21
## 3 417EE4… electr… member 398 0 2021-01-14 2021 01 Thursd… 17
## 4 FBFC52… electr… member 455 0 2021-01-14 2021 01 Thursd… 17
## 5 578B5E… electr… casual 283 0 2021-02-24 2021 02 Wednes… 21
## 6 2A63B4… classi… member 0 0 2021-02-08 2021 02 Monday 11
## # … with abbreviated variable names ¹rideable_type, ²member_casual,
## # ³ride_distance, ⁴ride_length, ⁵day_of_week, ⁶hour_of_day
Here is a summary of the ride length column (all figures in seconds)
summary(df_v2$ride_length)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0 390 695 1184 1261 3356649
The overall ride length average is
as.integer(mean(df_v2$ride_length)/12)) min
aggregate(df_v2$ride_length ~ df_v2$member_casual, FUN = mean)## df_v2$member_casual df_v2$ride_length
## 1 casual 1685.6687
## 2 member 787.5082
The average ride length of casual riders is significantly greater than the average ride length of annual members
One way to explain this is that annual members most likely use bikes to go to work, whereas casual riders may use them more for touristic purpose.
aggregate(df_v2$ride_length ~ df_v2$member_casual, FUN = median)## df_v2$member_casual df_v2$ride_length
## 1 casual 920
## 2 member 564
aggregate(df_v2$ride_length ~ df_v2$member_casual, FUN = max)## df_v2$member_casual df_v2$ride_length
## 1 casual 3356649
## 2 member 89996
aggregate(df_v2$ride_length ~ df_v2$member_casual, FUN = min)## df_v2$member_casual df_v2$ride_length
## 1 casual 0
## 2 member 0
The median ride length of casual riders is significantly greater than the median ride length of annual members
# Notice that the days of the week are out of order. Let's fix that.
df_v2$day_of_week <- ordered(df_v2$day_of_week, levels=c("Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday","Sunday"))
# Now, let's run the average ride time by each day for members vs casual users
aggregate(df_v2$ride_length ~ df_v2$member_casual + df_v2$day_of_week, FUN = mean)## df_v2$member_casual df_v2$day_of_week df_v2$ride_length
## 1 casual Monday 1708.4593
## 2 member Monday 764.8715
## 3 casual Tuesday 1491.0370
## 4 member Tuesday 742.2178
## 5 casual Wednesday 1460.9048
## 6 member Wednesday 744.3941
## 7 casual Thursday 1461.2255
## 8 member Thursday 746.8582
## 9 casual Friday 1582.8132
## 10 member Friday 767.7987
## 11 casual Saturday 1844.1079
## 12 member Saturday 880.4036
## 13 casual Sunday 1953.5458
## 14 member Sunday 890.1402
df_v2 %>%
group_by(member_casual) %>%
summarise(average_distance = mean(ride_distance)) %>%
arrange(member_casual) %>%
ggplot(aes(x = member_casual, y = average_distance, fill = member_casual)) +geom_col()
df_v2 %>%
group_by(member_casual, day_of_week) %>%
summarise(number_of_rides = n(),average_duration = mean(ride_length)) %>%
arrange(member_casual, day_of_week) %>%
ggplot(aes(x = day_of_week, y = average_duration, fill = member_casual)) +geom_col(position = "dodge")## `summarise()` has grouped output by 'member_casual'. You can override using the
## `.groups` argument.
df_v2 %>%
group_by(member_casual, day_of_week) %>%
summarise(number_of_rides = n(),average_duration = mean(ride_length)) %>%
arrange(member_casual, day_of_week) %>%
ggplot(aes(x = day_of_week, y = number_of_rides, fill = member_casual)) +geom_col(position = "dodge")## `summarise()` has grouped output by 'member_casual'. You can override using the
## `.groups` argument.
df_v2 %>%
group_by(member_casual, month)%>%
summarise(number_of_rides = n(),average_duration = mean(ride_length))%>%
arrange(member_casual, month)%>%
ggplot(aes(x = month, y = average_duration, fill = member_casual)) +geom_col(position = "dodge")## `summarise()` has grouped output by 'member_casual'. You can override using the
## `.groups` argument.
df_v2 %>%
group_by(member_casual, month)%>%
summarise(number_of_rides = n(),average_duration = mean(ride_length))%>%
arrange(member_casual, month)%>%
ggplot(aes(x = month, y = number_of_rides, fill = member_casual)) +geom_col(position = "dodge")## `summarise()` has grouped output by 'member_casual'. You can override using the
## `.groups` argument.
df_v2 %>%
group_by(member_casual, date)%>%
summarise(number_of_rides = n(),average_duration = mean(ride_length))%>%
arrange(member_casual, date)%>%
ggplot(aes(x = date, y = number_of_rides, color = member_casual)) +geom_line()## `summarise()` has grouped output by 'member_casual'. You can override using the
## `.groups` argument.
df_v2 %>%
group_by(member_casual, hour_of_day)%>%
summarise(number_of_rides = n(),average_duration = mean(ride_length))%>%
arrange(member_casual, hour_of_day)%>%
ggplot(aes(x = hour_of_day, y = number_of_rides, color = member_casual)) +geom_line()## `summarise()` has grouped output by 'member_casual'. You can override using the
## `.groups` argument.
The bike trip number of rides has a yearly seasonality which can be explained by the weather: people use less often bikes in the winter in Chicago, no matter if it is an annual member or a casual rider. However, the casual riders number of rides show a weekly seanality whereas it is not the case for annual members ; annual members tend to use their bikes the same way throughout the week, while casual riders more oftenly use bikes on the weekend.
Scarcely ride Cyclistic bikes during the colder months
Pick up and drop off their bikes at roughly the same distance
Are more numerous during summer and congregate near the bay area
Tend to be more active during the afternoon and late evenings
Consistently have longer ride lengths year-round
Favor docked bikes for lengthy bike trips
Have a negligible presence during the winter months
Have steady average ride lengths year-round
Are active at all times of the day, more so in the morning and
afternoon
Have trips spread out throughout Chicago’s downtown area
Are more consistent with their bike trips year-round
We make that assumption based on the fact that casuals:
Bike twice as much on Saturdays and Sundays compared to any other day of
the week
Spend the majority of their bike trips near parks and water
Spend significantly longer on average on every bike trip, suggesting
that they spend time in-between docking stations doing leisurely
activities
Scarcely use Cyclistic bikes in the morning (6 am-noon)
Do not use Cyclistic bikes often enough to warrant paying for an annual
membership
Bbased on the fact that members:
Rely on bikes consistently each week and year-round, with no notable
preference on a single day of the week
Use Cyclistic bikes often during the rush hours on a typical workday
Have a large geographical spread in the downtown area, particularly in
high urban dense areas
Are motivated by the economics of an annual membership pass
Age, gender. do we target the right people, potential to be more inclusive? Pricing plan, to understand spending behavior of casual users