Cyclistic bike share analysis
Carlos Vasquez
09/20/2021
===================== # STEP 1: Set up my environment # =====================
Load library packages and upload the previous 12 months (from time of date,9/20/2021) divvy-tripdata sets.
library(tidyverse)
library(janitor)
library(lubridate)
library(scales)q9_2020 <- read_csv("202009-divvy-tripdata.csv")
q10_2020 <- read_csv("202010-divvy-tripdata.csv")
q11_2020 <- read_csv("202011-divvy-tripdata.csv")
q12_2020 <- read_csv("202012-divvy-tripdata.csv")
q1_2021 <- read_csv("202101-divvy-tripdata.csv")
q2_2021 <- read_csv("202102-divvy-tripdata.csv")
q3_2021 <- read_csv("202103-divvy-tripdata.csv")
q4_2021 <- read_csv("202104-divvy-tripdata.csv")
q5_2021 <- read_csv("202105-divvy-tripdata.csv")
q6_2021 <- read_csv("202106-divvy-tripdata.csv")
q7_2021 <- read_csv("202107-divvy-tripdata.csv")
q8_2021 <- read_csv("202108-divvy-tripdata.csv")===================== # STEP 2. Make columns consistent and merge them into a single dataframe. # =====================
Use colnames function to compare the column names of each data set
#Note all column names were the same but I was unable to merge.
colnames(q9_2020)## [1] "ride_id" "rideable_type" "started_at"
## [4] "ended_at" "start_station_name" "start_station_id"
## [7] "end_station_name" "end_station_id" "start_lat"
## [10] "start_lng" "end_lat" "end_lng"
## [13] "member_casual"#colnames(q10_2020)
#colnames(q11_2020)
#colnames(q12_2020)
#colnames(q1_2021)
#colnames(q2_2021)
#colnames(q3_2021)
#colnames(q4_2021)
#colnames(q5_2021)
#colnames(q6_2021)
#colnames(q7_2021)
#colnames(q8_2021)Look for inconsistent data types
#inconsistent data type
sapply(q9_2020,class)## $ride_id
## [1] "character"
##
## $rideable_type
## [1] "character"
##
## $started_at
## [1] "POSIXct" "POSIXt"
##
## $ended_at
## [1] "POSIXct" "POSIXt"
##
## $start_station_name
## [1] "character"
##
## $start_station_id
## [1] "numeric"
##
## $end_station_name
## [1] "character"
##
## $end_station_id
## [1] "numeric"
##
## $start_lat
## [1] "numeric"
##
## $start_lng
## [1] "numeric"
##
## $end_lat
## [1] "numeric"
##
## $end_lng
## [1] "numeric"
##
## $member_casual
## [1] "character"#inconsistent data type
sapply(q10_2020,class)## $ride_id
## [1] "character"
##
## $rideable_type
## [1] "character"
##
## $started_at
## [1] "POSIXct" "POSIXt"
##
## $ended_at
## [1] "POSIXct" "POSIXt"
##
## $start_station_name
## [1] "character"
##
## $start_station_id
## [1] "numeric"
##
## $end_station_name
## [1] "character"
##
## $end_station_id
## [1] "numeric"
##
## $start_lat
## [1] "numeric"
##
## $start_lng
## [1] "numeric"
##
## $end_lat
## [1] "numeric"
##
## $end_lng
## [1] "numeric"
##
## $member_casual
## [1] "character"#inconsistent data type
sapply(q11_2020,class)## $ride_id
## [1] "character"
##
## $rideable_type
## [1] "character"
##
## $started_at
## [1] "POSIXct" "POSIXt"
##
## $ended_at
## [1] "POSIXct" "POSIXt"
##
## $start_station_name
## [1] "character"
##
## $start_station_id
## [1] "numeric"
##
## $end_station_name
## [1] "character"
##
## $end_station_id
## [1] "numeric"
##
## $start_lat
## [1] "numeric"
##
## $start_lng
## [1] "numeric"
##
## $end_lat
## [1] "numeric"
##
## $end_lng
## [1] "numeric"
##
## $member_casual
## [1] "character"#Observe start_station and end_station data type in a consistent data set
sapply(q12_2020,class)## $ride_id
## [1] "character"
##
## $rideable_type
## [1] "character"
##
## $started_at
## [1] "POSIXct" "POSIXt"
##
## $ended_at
## [1] "POSIXct" "POSIXt"
##
## $start_station_name
## [1] "character"
##
## $start_station_id
## [1] "character"
##
## $end_station_name
## [1] "character"
##
## $end_station_id
## [1] "character"
##
## $start_lat
## [1] "numeric"
##
## $start_lng
## [1] "numeric"
##
## $end_lat
## [1] "numeric"
##
## $end_lng
## [1] "numeric"
##
## $member_casual
## [1] "character"#consistent data sets
#sapply(q1_2021,class)
#sapply(q2_2021,class)
#sapply(q3_2021,class)
#sapply(q4_2021,class)
#sapply(q5_2021,class)
#sapply(q6_2021,class)
#sapply(q7_2021,class)
#sapply(q8_2021,class)Mutate data type to make all columns consistent for merging
q9_2020 <- mutate(q9_2020, start_station_id = as.character(start_station_id))
q10_2020 <- mutate(q10_2020, start_station_id = as.character(start_station_id))
q11_2020 <- mutate(q11_2020, start_station_id = as.character(start_station_id))
q9_2020 <- mutate(q9_2020, end_station_id = as.character(end_station_id))
q10_2020 <- mutate(q10_2020, end_station_id = as.character(end_station_id))
q11_2020 <- mutate(q11_2020, end_station_id = as.character(end_station_id))Merge into one data frame
bike_rides <- bind_rows(q9_2020, q10_2020, q11_2020, q12_2020, q1_2021, q2_2021, q3_2021, q4_2021, q5_2021, q6_2021, q7_2021, q8_2021)===================== # STEP 3. Prepare data for analysis # =====================
Inspect the new data frame
dim(bike_rides)## [1] 4913072 13Create minutes (ride length) column by subtracting ended_at column from started_at column.
bike_rides$minutes <- difftime(bike_rides$ended_at,bike_rides$started_at,units = c("min"))
bike_rides$minutes <- as.numeric(as.character(bike_rides$minutes))
bike_rides$minutes <- round(bike_rides$minutes, digits = 1)#round to tenth decimal placeCreate columns for: month, day, year, day of week, and hour.
bike_rides$date <- as.Date(bike_rides$started_at)
bike_rides$month <- format(as.Date(bike_rides$date), "%m")
bike_rides$day <- format(as.Date(bike_rides$date), "%d")
bike_rides$year <- format(as.Date(bike_rides$date), "%Y")
bike_rides$day_of_week <- format(as.Date(bike_rides$date), "%A")
bike_rides$hour <- lubridate::hour(bike_rides$started_at)Double check newly converted data types
is.numeric(bike_rides$minutes)## [1] TRUEis.Date(bike_rides$date)## [1] TRUEUse mutate function to create: season (Spring, Summer, Fall, Winter) column
bike_rides <-bike_rides %>% mutate(season =
case_when(month == "03" ~ "Spring",
month == "04" ~ "Spring",
month == "05" ~ "Spring",
month == "06" ~ "Summer",
month == "07" ~ "Summer",
month == "08" ~ "Summer",
month == "09" ~ "Fall",
month == "10" ~ "Fall",
month == "11" ~ "Fall",
month == "12" ~ "Winter",
month == "01" ~ "Winter",
month == "02" ~ "Winter"))time_of_day (Night, Morning, Afternoon, Evening,) and
bike_rides <-bike_rides %>% mutate(time_of_day =
case_when(hour == "0" ~ "Night",
hour == "1" ~ "Night",
hour == "2" ~ "Night",
hour == "3" ~ "Night",
hour == "4" ~ "Night",
hour == "5" ~ "Night",
hour == "6" ~ "Morning",
hour == "7" ~ "Morning",
hour == "8" ~ "Morning",
hour == "9" ~ "Morning",
hour == "10" ~ "Morning",
hour == "11" ~ "Morning",
hour == "12" ~ "Afternoon",
hour == "13" ~ "Afternoon",
hour == "14" ~ "Afternoon",
hour == "15" ~ "Afternoon",
hour == "16" ~ "Afternoon",
hour == "17" ~ "Afternoon",
hour == "18" ~ "Evening",
hour == "19" ~ "Evening",
hour == "20" ~ "Evening",
hour == "21" ~ "Evening",
hour == "22" ~ "Evening",
hour == "23" ~ "Evening"))to mutate the month column to display the full month name.
bike_rides <-bike_rides %>% mutate(month = case_when(month == "01" ~ "January",
month == "02" ~ "February",
month == "03" ~ "March",
month == "04" ~ "April",
month == "05" ~ "May",
month == "06" ~ "June",
month == "07" ~ "July",
month == "08" ~ "August",
month == "09" ~ "September",
month == "10" ~ "October",
month == "11" ~ "November",
month == "12" ~ "December"))===================== # STEP 5. Clean the data # =====================
Note: Business task: How do annual members and casual riders use Cyclistic bikes differently? Since our analyses is focusing on casual vs member riders let ensure our data reflects this.
unique(bike_rides$member_casual)## [1] "casual" "member"Remove empty columns, rows and remove NA values all into a new data frame
df <- janitor::remove_empty(bike_rides, which = c("cols"))
df <- janitor::remove_empty(bike_rides, which = c("rows"))
df <- distinct(bike_rides)
df<- na.omit(bike_rides)View the dimension
dim(df)## [1] 4233298 22Note: Number of observations is now 4,233,298 (679,774 rows were removed). Now filter the data frame to remove where ride_length is 0 or negative and filter out unnecessary columns.
df <- df %>%
filter(minutes>0) %>%
select(-c(ride_id,started_at,ended_at,start_station_id,end_station_name,end_station_id,start_lat,start_lng,end_lat,end_lng))Note: New data frame is 4,221,509 observations (11,789 additional observations were removed). View the final data frame.
View(df)
dim(df)## [1] 4221509 12===================== # STEP 5. Conduct descriptive analysis # =====================
Business task: How do annual members and casual riders use Cyclistic bikes differently?
Casual = customers who purchase single-ride or full-day passes
Members = customers who purchase annual memberships
What date range does our data cover?
## [1] "2020-09-01"to
## [1] "2021-08-31"How many total rides?
## [1] 4221509Find the number of rides per month
## # A tibble: 12 x 3
## # Groups: month [12]
## month year n
## <fct> <chr> <int>
## 1 September 2020 497294
## 2 October 2020 336698
## 3 November 2020 221591
## 4 December 2020 113371
## 5 January 2021 83366
## 6 February 2021 42840
## 7 March 2021 205454
## 8 April 2021 297801
## 9 May 2021 450278
## 10 June 2021 607945
## 11 July 2021 691376
## 12 August 2021 673495Lets visualize the data.
Our data covers 12 months, 2020-09-01 to 2021-08-31, that is the end of 2020 to the beginning of 2021. Lets visualize our graph chronologically. Image 2
Viewing the data in chronological order by month makes the data into a bell shape distribution. We can see that the peak of bike rides takes place in the month of July. For sake of this analysis, the season will be as follows. 
The peak months of number of bike rides are in the months of June-August,summer time. We will come back to this time frame. Is there a difference between type of riders and number of rides in the overall data? 
At hindsight we can see the number of bike rides for both member and casual riders are at its highest levels during the summer time (June-August). The total number of rides during summer time is
## # A tibble: 1 x 2
## n prop
## <int> <dbl>
## 1 1972816 0.467Around 47 percent of all rides take place during the summer time. Let’s focus and continue our analysis in this time frame (June-August). First lets find the total number of riders by type of rider.
## # A tibble: 2 x 3
## member_casual n prop
## <chr> <int> <dbl>
## 1 casual 1014122 0.514
## 2 member 958694 0.486During summer time, casual riders tend to out number the member riders. As shown above (Image 2), July was the busiest month with casual riders outnumbering members during July. What are the figures of the type of bicycle used during June-August?
## # A tibble: 3 x 3
## rideable_type n prop
## <chr> <int> <dbl>
## 1 classic_bike 1439012 0.729
## 2 docked_bike 154390 0.0783
## 3 electric_bike 379414 0.192The most popular bikes during June-August was classic bikes.Users used classic bikes 9.3 more times than docked bikes and 3.8 more times than electric bikes. The individual numbers by month and type of bike are as follows:
## # A tibble: 9 x 4
## month rideable_type n prop
## <fct> <chr> <int> <dbl>
## 1 June classic_bike 433145 0.220
## 2 June docked_bike 51694 0.0262
## 3 June electric_bike 123106 0.0624
## 4 July classic_bike 504791 0.256
## 5 July docked_bike 57664 0.0292
## 6 July electric_bike 128921 0.0653
## 7 August classic_bike 501076 0.254
## 8 August docked_bike 45032 0.0228
## 9 August electric_bike 127387 0.0646Lets visualize and lets also consider the type of member utilizing these bikes during the summer. 
As mentioned earlier, users use classic bikes 9.3 more times than docked bikes and 3.8 more times than electric bikes. Classic bikes are favorable regardless of type of rider and summer month. Individual number of graphs are below:
## # A tibble: 15 x 4
## month rideable_type member_casual n
## <fct> <chr> <chr> <int>
## 1 June classic_bike casual 187234
## 2 June docked_bike casual 51694
## 3 June electric_bike casual 64976
## 4 July classic_bike casual 240315
## 5 July docked_bike casual 57664
## 6 July electric_bike casual 71073
## 7 August classic_bike casual 228931
## 8 August docked_bike casual 45032
## 9 August electric_bike casual 67203
## 10 June classic_bike member 245911
## 11 June electric_bike member 58130
## 12 July classic_bike member 264476
## 13 July electric_bike member 57848
## 14 August classic_bike member 272145
## 15 August electric_bike member 60184Lets find the mean, median, max, and min for the ride length (minutes) for customers during summer time.
## # A tibble: 1 x 4
## Average_ride_length min med max
## <dbl> <dbl> <dbl> <dbl>
## 1 23.8 0.1 13.3 55944.Between casual riders and members.
## # A tibble: 2 x 5
## member_casual Average_duration min med max
## <chr> <dbl> <dbl> <dbl> <dbl>
## 1 casual 33.3 0.1 17.2 55944.
## 2 member 13.8 0.1 10.4 1496.Not only do casual riders outnumber members they also on average spend longer time riding bicycles than members. What are the average ride length between casual rider and members in a a given day ? (Note: Order the days of the week to make it easy to analyse.)
df$day_of_week <- ordered(df$day_of_week, levels=c("Sunday", "Monday","Tuesday", "Wednesday", "Thursday", "Friday", "Saturday"))Find the average minutes spend riding bikes by day of the week between casual riders and members.
## `summarise()` has grouped output by 'member_casual'. You can override using the `.groups` argument.## # A tibble: 14 x 3
## # Groups: member_casual [2]
## member_casual day_of_week average_duration
## <chr> <ord> <dbl>
## 1 casual Sunday 37.2
## 2 casual Monday 32.4
## 3 casual Tuesday 29.6
## 4 casual Wednesday 30.4
## 5 casual Thursday 30.9
## 6 casual Friday 31.8
## 7 casual Saturday 35.9
## 8 member Sunday 15.8
## 9 member Monday 13.2
## 10 member Tuesday 13.1
## 11 member Wednesday 13.2
## 12 member Thursday 13.2
## 13 member Friday 13.5
## 14 member Saturday 15.5Lets visualize (Note: Visualization is comparing casual riders vs members).
Casual = customers who purchase single-ride or full-day passes
Members = customers who purchase annual memberships
## `summarise()` has grouped output by 'member_casual'. You can override using the `.groups` argument.
Is there a change when we filter for type of bike used?
## `summarise()` has grouped output by 'member_casual', 'day_of_week'. You can override using the `.groups` argument.
Look at this. Casual riders on average spend more time riding docked bikes on any given day of the week.
## `summarise()` has grouped output by 'member_casual'. You can override using the `.groups` argument.## # A tibble: 5 x 3
## # Groups: member_casual [2]
## member_casual rideable_type average_duration
## <chr> <chr> <dbl>
## 1 casual classic_bike 25.9
## 2 casual docked_bike 80.4
## 3 casual electric_bike 21.3
## 4 member classic_bike 14.1
## 5 member electric_bike 12.9Casual riders spend on average 3.1 times longer riding docked bicycles compared with classic bicycles. We will come back to this. For now lets find the number of rides per day of the week between casual riders and members
## `summarise()` has grouped output by 'member_casual'. You can override using the `.groups` argument.## # A tibble: 14 x 4
## # Groups: member_casual [2]
## member_casual day_of_week number_of_rides average_duration
## <chr> <ord> <int> <dbl>
## 1 casual Sunday 191607 37.2
## 2 casual Monday 108241 32.4
## 3 casual Tuesday 112901 29.6
## 4 casual Wednesday 109301 30.4
## 5 casual Thursday 117835 30.9
## 6 casual Friday 150376 31.8
## 7 casual Saturday 223861 35.9
## 8 member Sunday 119107 15.8
## 9 member Monday 128107 13.2
## 10 member Tuesday 151194 13.1
## 11 member Wednesday 145784 13.2
## 12 member Thursday 143466 13.2
## 13 member Friday 138681 13.5
## 14 member Saturday 132355 15.5Visualize the number of rides by rider type
## `summarise()` has grouped output by 'member_casual'. You can override using the `.groups` argument.
Lets see the difference between the number of rider per day by analyzing by type of bike
## `summarise()` has grouped output by 'member_casual', 'day_of_week'. You can override using the `.groups` argument.
Even though casual riders on average spend more time riding docked bikes on any given day of the week, docked bicycles are not used as frequently compared to classic and electric bicycles.
## # A tibble: 5 x 4
## rideable_type member_casual n prop
## <chr> <chr> <int> <dbl>
## 1 classic_bike casual 656480 0.333
## 2 classic_bike member 782532 0.397
## 3 docked_bike casual 154390 0.0783
## 4 electric_bike casual 203252 0.103
## 5 electric_bike member 176162 0.0893Casual riders use classic bicycles 4.3 more times than docked bicycles. What time during the day do we see the most riders?
## # A tibble: 48 x 3
## # Groups: member_casual [2]
## member_casual hour n
## <chr> <int> <int>
## 1 member 17 104359
## 2 casual 17 95257
## 3 member 18 90221
## 4 casual 18 89295
## 5 casual 16 78423
## 6 member 16 77755
## 7 casual 19 73276
## 8 casual 15 70212
## 9 member 19 68617
## 10 casual 14 67790
## # ... with 38 more rowsLets visualize
Visualize for time of day and during the day of the week between casual riders and members. 
The afternoon is the peak time the most riders come on any given day of the week. Casual drivers come most on Saturday and Sunday. Popular Start Stations for Casual riders are:
## `summarise()` has grouped output by 'member_casual'. You can override using the `.groups` argument.## Adding missing grouping variables: `member_casual`## # A tibble: 30 x 3
## # Groups: member_casual [1]
## member_casual start_station_name number_of_ride
## <chr> <chr> <int>
## 1 casual Streeter Dr & Grand Ave 36421
## 2 casual Michigan Ave & Oak St 16113
## 3 casual Millennium Park 15963
## 4 casual Theater on the Lake 11798
## 5 casual Shedd Aquarium 11218
## 6 casual Wells St & Concord Ln 9804
## 7 casual Lake Shore Dr & North Blvd 9546
## 8 casual Lake Shore Dr & Monroe St 9383
## 9 casual Clark St & Lincoln Ave 8697
## 10 casual DuSable Lake Shore Dr & North Blvd 8273
## # ... with 20 more rowsPopular Start Stations for Member riders:
## `summarise()` has grouped output by 'member_casual'. You can override using the `.groups` argument.## Adding missing grouping variables: `member_casual`## # A tibble: 30 x 3
## # Groups: member_casual [1]
## member_casual start_station_name number_of_ride
## <chr> <chr> <int>
## 1 member Wells St & Concord Ln 9337
## 2 member Clark St & Elm St 9097
## 3 member Kingsbury St & Kinzie St 8197
## 4 member Streeter Dr & Grand Ave 7864
## 5 member Wells St & Elm St 7858
## 6 member Theater on the Lake 7465
## 7 member Clark St & Lincoln Ave 7044
## 8 member Michigan Ave & Oak St 6782
## 9 member Broadway & Barry Ave 6739
## 10 member Wells St & Huron St 6727
## # ... with 20 more rowsEnd of analysis.
Summary:
-I learned that docked bicycle type is on average ridden longer by casual riders. However, casual riders use classic bicycles 4.3 more than docked bicycles.
-Saturday and Sunday afternoons are the most popular riding days for casual riders.
-November through February have the least number of casual riders while June, July, and August have a particularly high number of Casual riders.
-The most popular stations for Casual riders in descending order are Streeter Dr & Grand Ave, Michigan Ave & Oak St, Millennium Park, Theater on the Lake, Shedd Aquarium.
Recommendations
-Based on the data analyzed I would recommend we focus our marketing efforts for Casual riders with these parameters
1: Increase marketing for docket bicycles 2. Heavier marketing from June through August 3. Focus marketing on afternoon weekends 4. Invest in marketing at the top 5 stations as noted above.