Diego 2023-10-16
This analysis is part of a Capstone Project for the Google Data Analytics Certificate program (Cyclistic). It will use the software RStudio and R programming language to complete the project, and for sharing the results, this notebook markdown, and a Slide presentation will be created.
The project will be completed by using the 6 Data Analytics stages:
- Ask: Identify the business task and determine the key stakeholders.
- Prepare: Collect the data, identify how it’s organized, determine the credibility of the data.
- Process: Select the tool for data cleaning, check for errors and document the cleaning process.
- Analyze: Organize and format the data, aggregate the data so that it’s useful, perform calculations and identify trends and relationships.
- Share: Use design thinking principles and data-driven storytelling approach, present the findings with effective visualization. Ensure the analysis has answered the business task.
- Act: Share the final conclusion and the recommendations.
Cyclistic, a bike-share program in Chicago, started in 2016 and has since grown to operate a fleet of 5,824 bikes across 692 geotracked stations. Cyclistic offers flexible pricing plans, including single-ride passes, full-day passes, and annual memberships. The annual members have been found to be more profitable than casual riders, and the company aims to increase the number of annual members.
To achieve this goal, Cyclistic’s marketing team, led by Moreno, plans to convert casual riders into annual members. Moreno believes that casual riders, who are already aware of Cyclistic, can be persuaded to become members. The team intends to analyze historical bike trip data to understand the differences between annual members and casual riders, why casual riders might choose a membership, and how digital media can be used to enhance their marketing strategies. This data-driven approach will help Cyclistic develop effective marketing strategies for achieving their goal.
-
Guiding questions
- What is the problem you are trying to solve?
- How can your insights drive business decisions?
-
Key tasks
- Identify the business task
- Consider key stakeholders
-
Deliverable
- A clear statement of the business task
How do annual members and casual riders use Cyclistic bikes differently?
By completing the business task Cyclistic marketing team will be able to achieve it’s business goal of design marketing strategies aimed at converting casual riders into annual members.
- Cyclistic: A bike-share program that features more than 5,800 bicycles and 600 docking stations. Cyclistic sets itself apart by also offering reclining bikes, hand tricycles, and cargo bikes, making bike-share more inclusive to people with disabilities and riders who can’t use a standard two-wheeled bike. The majority of riders opt for traditional bikes; about 8% of riders use the assistive options. Cyclistic users are more likely to ride for leisure, but about 30% use them to commute to work each day.
- Lily Moreno: The director of marketing. Moreno is responsible for the development of campaigns and initiatives to promote the bike-share program. These may include email, social media, and other channels.
- Cyclistic marketing analytics team: A team of data analysts who are responsible for collecting, analyzing, and reporting data that helps guide Cyclistic marketing strategy.
- Cyclistic executive team: The notoriously detail-oriented executive team will decide whether to approve the recommended marketing program.
-
Guiding questions
- Where is your data located?
- How is the data organized?
- Are there issues with bias or credibility in this data? Does your data ROCCC(reliable, original, comprehensive, current and cited)?
- How are you addressing licensing, privacy, security, and accessibility?
- How did you verify the data’s integrity?
- How does it help you answer your question?
- Are there any problems with the data?
-
Key tasks
- Download data and store it appropriately.
- Identify how it’s organized.
- Sort and filter the data.
- Determine the credibility of the data.
-
Deliverable
- A description of all data sources used
The data used for this project is located at this Data Source and has been made available by Motivate International Inc. under this License. There are various compressed files in .zip format, with variations based on the year of the information.
For this analysis it will be used the past 12 months of data files from 08/2022 to 07/2023. Below is the list of downloaded files.
- 202208-divvy-tripdata.zip
- 202209-divvy-tripdata.zip
- 202210-divvy-tripdata.zip
- 202211-divvy-tripdata.zip
- 202212-divvy-tripdata.zip
- 202301-divvy-tripdata.zip
- 202302-divvy-tripdata.zip
- 202303-divvy-tripdata.zip
- 202304-divvy-tripdata.zip
- 202305-divvy-tripdata.zip
- 202306-divvy-tripdata.zip
- 202307-divvy-tripdata.zip
The extracted files are in .csv (comma-separated values) format, organized by month, starting from August 2022 and ending in July 2023, totaling 12 CSV files. To check these files, the R programming language will be used.
# Install the 'tidyverse' package, a collection of R packages for data manipulation and visualization.
install.packages("tidyverse", repos = "https://cran.rstudio.com/")##
## The downloaded binary packages are in
## /var/folders/vl/1wy31mz11vgfjtq8g4t2slzc0000gn/T//RtmpxfyvIe/downloaded_packages
# Install the 'data.table' package, a high-performance data manipulation package.
install.packages("data.table", repos = "https://cran.rstudio.com/")##
## The downloaded binary packages are in
## /var/folders/vl/1wy31mz11vgfjtq8g4t2slzc0000gn/T//RtmpxfyvIe/downloaded_packages
# Install the 'osmdata' package for working with OpenStreetMap data.
install.packages("osmdata", repos = "https://cran.rstudio.com/")##
## The downloaded binary packages are in
## /var/folders/vl/1wy31mz11vgfjtq8g4t2slzc0000gn/T//RtmpxfyvIe/downloaded_packages
# Load packages into R environment.
library("tidyverse")## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr 1.1.3 ✔ readr 2.1.4
## ✔ forcats 1.0.0 ✔ stringr 1.5.0
## ✔ ggplot2 3.4.3 ✔ tibble 3.2.1
## ✔ lubridate 1.9.2 ✔ tidyr 1.3.0
## ✔ purrr 1.0.2
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
library("data.table")##
## Attaching package: 'data.table'
##
## The following objects are masked from 'package:lubridate':
##
## hour, isoweek, mday, minute, month, quarter, second, wday, week,
## yday, year
##
## The following objects are masked from 'package:dplyr':
##
## between, first, last
##
## The following object is masked from 'package:purrr':
##
## transpose
library("osmdata")## Data (c) OpenStreetMap contributors, ODbL 1.0. https://www.openstreetmap.org/copyright
# Load additional packages for data manipulation and analysis.
library("dplyr")
library("lubridate")
library("janitor")##
## Attaching package: 'janitor'
##
## The following objects are masked from 'package:stats':
##
## chisq.test, fisher.test
# Load 'ggplot2' for data visualization.
library("ggplot2")
# Load 'ggmap' for integrating Google Maps with R.
library("ggmap")## The legacy packages maptools, rgdal, and rgeos, underpinning the sp package,
## which was just loaded, were retired in October 2023.
## Please refer to R-spatial evolution reports for details, especially
## https://r-spatial.org/r/2023/05/15/evolution4.html.
## It may be desirable to make the sf package available;
## package maintainers should consider adding sf to Suggests:.
## ℹ Google's Terms of Service: <https://mapsplatform.google.com>
## ℹ Please cite ggmap if you use it! Use `citation("ggmap")` for details.
# Create data frames from each CSV file.
df_202208 <- fread("202208-divvy-tripdata.csv")
df_202209 <- fread("202209-divvy-publictripdata.csv")
df_202210 <- fread("202210-divvy-tripdata.csv")
df_202211 <- fread("202211-divvy-tripdata.csv")
df_202212 <- fread("202212-divvy-tripdata.csv")
df_202301 <- fread("202301-divvy-tripdata.csv")
df_202302 <- fread("202302-divvy-tripdata.csv")
df_202303 <- fread("202303-divvy-tripdata.csv")
df_202304 <- fread("202304-divvy-tripdata.csv")
df_202305 <- fread("202305-divvy-tripdata.csv")
df_202306 <- fread("202306-divvy-tripdata.csv")
df_202307 <- fread("202307-divvy-tripdata.csv")To gain an initial understanding of the data, use the str() function, which provides a concise summary. Subsequently, the goal is to consolidate all the data frames into a single unified data frame. To achieve this, it is imperative that all data frames share identical column names and data types.
Once the data has been merged into one comprehensive data frame, the next crucial step is to identify duplicates, NA (Not Available), NaN (Not-A-Number), and empty values.
str(df_202307)## Classes 'data.table' and 'data.frame': 767650 obs. of 13 variables:
## $ ride_id : chr "9340B064F0AEE130" "D1460EE3CE0D8AF8" "DF41BE31B895A25E" "9624A293749EF703" ...
## $ rideable_type : chr "electric_bike" "classic_bike" "classic_bike" "electric_bike" ...
## $ started_at : POSIXct, format: "2023-07-23 20:06:14" "2023-07-23 17:05:07" ...
## $ ended_at : POSIXct, format: "2023-07-23 20:22:44" "2023-07-23 17:18:37" ...
## $ start_station_name: chr "Kedzie Ave & 110th St" "Western Ave & Walton St" "Western Ave & Walton St" "Racine Ave & Randolph St" ...
## $ start_station_id : chr "20204" "KA1504000103" "KA1504000103" "13155" ...
## $ end_station_name : chr "Public Rack - Racine Ave & 109th Pl" "Milwaukee Ave & Grand Ave" "Damen Ave & Pierce Ave" "Clinton St & Madison St" ...
## $ end_station_id : chr "877" "13033" "TA1305000041" "TA1305000032" ...
## $ start_lat : num 41.7 41.9 41.9 41.9 42 ...
## $ start_lng : num -87.7 -87.7 -87.7 -87.7 -87.7 ...
## $ end_lat : num 41.7 41.9 41.9 41.9 42 ...
## $ end_lng : num -87.7 -87.6 -87.7 -87.6 -87.6 ...
## $ member_casual : chr "member" "member" "member" "member" ...
## - attr(*, ".internal.selfref")=<externalptr>
# Create a list of data frames.
list_of_dfs <- list(
df_202208, df_202209, df_202210, df_202211, df_202212, df_202301, df_202302,
df_202303, df_202304, df_202305, df_202306, df_202307
)
# Compare data frames columns and return "TRUE" if they can be bound together
# or "FALSE" along with a list indicating where the columns are different.
compare_df_cols_same(list_of_dfs)## [1] TRUE
# Joining all data frames in one.
data <- rbindlist(list_of_dfs)
# Check for NA values.
colSums(is.na(data))## ride_id rideable_type started_at ended_at
## 0 0 0 0
## start_station_name start_station_id end_station_name end_station_id
## 0 0 0 0
## start_lat start_lng end_lat end_lng
## 0 0 6102 6102
## member_casual
## 0
# Check for NAN values.
colSums(sapply(data, is.nan))## ride_id rideable_type started_at ended_at
## 0 0 0 0
## start_station_name start_station_id end_station_name end_station_id
## 0 0 0 0
## start_lat start_lng end_lat end_lng
## 0 0 0 0
## member_casual
## 0
# Check for empty values in the data, exclude columns "started_at" and "ended_at".
# It will be possible to check for empty values in these columns further.
colSums(data[, !c("started_at", "ended_at")] == "")## ride_id rideable_type start_station_name start_station_id
## 0 0 868772 868904
## end_station_name end_station_id start_lat start_lng
## 925008 925149 0 0
## end_lat end_lng member_casual
## NA NA 0
# Check for duplicates in the only possible column that must not have it.
sum(duplicated(data$ride_id))## [1] 0
summary(data)## ride_id rideable_type started_at
## Length:5723606 Length:5723606 Min. :2022-08-01 00:00:00.00
## Class :character Class :character 1st Qu.:2022-09-28 13:56:43.50
## Mode :character Mode :character Median :2023-02-16 13:53:51.50
## Mean :2023-02-01 23:55:22.17
## 3rd Qu.:2023-06-03 07:41:37.00
## Max. :2023-07-31 23:59:56.00
##
## ended_at start_station_name start_station_id
## Min. :2022-08-01 00:05:00.00 Length:5723606 Length:5723606
## 1st Qu.:2022-09-28 14:12:20.25 Class :character Class :character
## Median :2023-02-16 14:04:56.50 Mode :character Mode :character
## Mean :2023-02-02 00:13:43.58
## 3rd Qu.:2023-06-03 08:00:15.00
## Max. :2023-08-12 04:53:41.00
##
## end_station_name end_station_id start_lat start_lng
## Length:5723606 Length:5723606 Min. :41.64 Min. :-87.92
## Class :character Class :character 1st Qu.:41.88 1st Qu.:-87.66
## Mode :character Mode :character Median :41.90 Median :-87.64
## Mean :41.90 Mean :-87.65
## 3rd Qu.:41.93 3rd Qu.:-87.63
## Max. :42.07 Max. :-87.52
##
## end_lat end_lng member_casual
## Min. : 0.00 Min. :-88.16 Length:5723606
## 1st Qu.:41.88 1st Qu.:-87.66 Class :character
## Median :41.90 Median :-87.64 Mode :character
## Mean :41.90 Mean :-87.65
## 3rd Qu.:41.93 3rd Qu.:-87.63
## Max. :42.18 Max. : 0.00
## NA's :6102 NA's :6102
-
From the last step “Data Check”:
-
NA values: Columns “end_lat” and “end_lng” missing values showed above indicate an error.
-
NAN values: None.
-
Empty cells: Columns “start_station_name”, “start_station_id”, “end_station_name” and “end_station_id” can affect data integrity.
-
Duplicates: None.
-
There is no metadata available, however, it is possible to identify its content by the name of the columns. Using the summary() and the str() function used before, it shows that there are over 4 million rows and 13 columns with the following names:
- ride_id #Ride id - unique
- rideable_type #Bike type - Classic, Docked, Electric
- started_at #Ride start day and time
- ended_at #Ride end day and time
- start_station_name #Ride start station name
- start_station_id #Ride start station id
- end_station_name #Ride end station name
- end_station_id #Ride end station id
- start_lat #Ride start latitude
- start_lng #Ride start longitute
- end_lat #Ride end latitude
- end_lng #Ride end longitute
- member_casual #Rider type - Member or Casual
- There is enough data to compare casual riders and members, a complete year, 12 months.
- While the data source contains historical data dating back to 2013, this analysis assumes that previous habits are unlikely to return after the COVID-19 pandemic. Additionally, the older files lack the same information, and any potentially valuable data is only available for members.
- Missing values and errors will be deleted before analysis.
-
Guiding questions
- What tools are you choosing and why?
- Have you ensured your data’s integrity?
- What steps have you taken to ensure that your data is clean?
- How can you verify that your data is clean and ready to analyze?
- Have you documented your cleaning process so you can review and share those results?
-
Key tasks
- Check the data for errors.
- Choose your tools.
- Transform the data so you can work with it effectively.
- Document the cleaning process.
-
Deliverable
- Documentation of any cleaning or manipulation of data
The data will be processed, analyzed and visualized using the R programming language.
First, it is necessary to remove missing values, and errors in latitude and longitude. Next, additional columns will be created for further analysis. Finally, the data set will be cleaned and prepared for analysis.
# Remove NA and empty values.
data <- na.omit(data)
data <- data %>% filter(if_all(starts_with("start_station_name"):ends_with("end_lng"), ~ . != ""))
# Get the bounding box coordinates for Chicago.
chicago_bb <- getbb("Chicago")
# Filter the data frame to include only rows within the Chicago bounding box.
data <- data %>%
filter(
start_lat >= chicago_bb[[2]] &
start_lat <= chicago_bb[[4]] &
start_lng >= chicago_bb[[1]] &
start_lng <= chicago_bb[[3]] &
end_lat >= chicago_bb[[2]] &
end_lat <= chicago_bb[[4]] &
end_lng >= chicago_bb[[1]] &
end_lng <= chicago_bb[[3]]
)
# Create useful variables for further analysis.
data <- data %>%
mutate(
date = as.Date(started_at),
day_of_week = weekdays(date),
month = months(date),
ride_time_secs = as.numeric(difftime(ended_at, started_at, units = "secs")),
period = case_when(
hour(started_at) >= 5 & hour(started_at) < 11 ~ 'Morning',
hour(started_at) >= 11 & hour(started_at) < 14 ~ 'Lunch',
hour(started_at) >= 14 & hour(started_at) < 18 ~ 'Afternoon',
hour(started_at) >= 18 & hour(started_at) < 22 ~ 'Evening',
hour(started_at) >= 22 | hour(started_at) < 5 ~ 'Night'),
season = case_when(
month(date) %in% c(12, 1, 2) ~ 'Winter',
month(date) %in% c(3, 4, 5) ~ 'Spring',
month(date) %in% c(6, 7, 8) ~ 'Summer',
month(date) %in% c(9, 10, 11) ~ 'Autumn')
)
# Data Check
summary(data)## ride_id rideable_type started_at
## Length:4308997 Length:4308997 Min. :2022-08-01 00:00:07.00
## Class :character Class :character 1st Qu.:2022-09-27 16:09:17.00
## Mode :character Mode :character Median :2023-02-15 08:14:46.00
## Mean :2023-02-01 08:12:32.14
## 3rd Qu.:2023-06-02 12:49:26.00
## Max. :2023-07-31 23:59:15.00
## ended_at start_station_name start_station_id
## Min. :2022-08-01 00:05:44.00 Length:4308997 Length:4308997
## 1st Qu.:2022-09-27 16:21:35.00 Class :character Class :character
## Median :2023-02-15 08:24:13.00 Mode :character Mode :character
## Mean :2023-02-01 08:28:24.11
## 3rd Qu.:2023-06-02 13:08:02.00
## Max. :2023-08-01 20:40:50.00
## end_station_name end_station_id start_lat start_lng
## Length:4308997 Length:4308997 Min. :41.65 Min. :-87.84
## Class :character Class :character 1st Qu.:41.88 1st Qu.:-87.66
## Mode :character Mode :character Median :41.90 Median :-87.64
## Mean :41.90 Mean :-87.64
## 3rd Qu.:41.93 3rd Qu.:-87.63
## Max. :42.02 Max. :-87.53
## end_lat end_lng member_casual date
## Min. :41.65 Min. :-87.84 Length:4308997 Min. :2022-08-01
## 1st Qu.:41.88 1st Qu.:-87.66 Class :character 1st Qu.:2022-09-27
## Median :41.90 Median :-87.64 Mode :character Median :2023-02-15
## Mean :41.90 Mean :-87.64 Mean :2023-01-31
## 3rd Qu.:41.93 3rd Qu.:-87.63 3rd Qu.:2023-06-02
## Max. :42.02 Max. :-87.53 Max. :2023-07-31
## day_of_week month ride_time_secs period
## Length:4308997 Length:4308997 Min. :-10122 Length:4308997
## Class :character Class :character 1st Qu.: 340 Class :character
## Mode :character Mode :character Median : 593 Mode :character
## Mean : 952
## 3rd Qu.: 1056
## Max. :728178
## season
## Length:4308997
## Class :character
## Mode :character
##
##
##
Based on the summary provided, there appear to be anomalies in the ‘ride_time_secs’ column. The minimum value should be greater than zero; any negative or “0” value is not a valid ride. While there is an exceptionally high maximum value, it may have occurred when the user returned the bike on a different day.
# Clean bad data.
data <- data %>% filter(ride_time_secs > 0)
# Create a clean data frame ready for analysis.
data_v2 <- data %>% select(-c(started_at, ended_at, start_station_id, end_station_id))
# Export the new data frame as a CSV file.
write.csv(data_v2, file = "cyclistic_202208_202307.csv")-
Data Cleaning & Manipulation Process:
-
Deletions:
- Rows with NA values in the “end_lat” and “end_lng” columns. These rides may be considered errors; as a hypothesis, the bikes might have been taken for maintenance.
- Rows with empty cells will be deleted to guarantee data integrity.
- Rows with latitudes and longitudes outside of Chicago.
- Rows with ride_time_secs <= 0. These represent invalid rides with negative or zero values.
- Columns “started_at”, “ended_at”, “start_station_id”, “end_station_id”. These columns will no longer be necessary.
-
Additions:
-
Date, Day of Week and Month Calculation:
-
date: Calculated by extracting the complete date from the started_at column.
-
day_of_week: Calculated by extracting the weekday from date column (e.g., “Monday”).
-
month: Calculated by extracting the month from date column (e.g., “January”).
-
-
Ride Time:
- ride_time_secs: Calculated as the numeric difference in seconds between ‘ended_at’ and ‘started_at’.
-
Period Classification:
- period: This column is determined based on the time of day in
the started_at column. Rides are categorized into different
periods, including “Morning,” “Lunch,” “Afternoon,” “Evening,”
and “Night,” based on the time of day.
- Morning: from 5 to 11
- Lunch: from 11 to 14
- Afternoon: from 14 to 18
- Evening: from 18 to 22
- Night: from 22 to 5
- period: This column is determined based on the time of day in
the started_at column. Rides are categorized into different
periods, including “Morning,” “Lunch,” “Afternoon,” “Evening,”
and “Night,” based on the time of day.
-
Season Classification:
- season: This column categorizes rides into different seasons
based on the month in the “started_at” column. The following
seasons are used:
- Winter: Rides occurring in December, January, or February.
- Spring: Rides occurring in March, April, or May.
- Summer: Rides occurring in June, July, or August.
- Autumn: Rides occurring in September, October, or November.
- season: This column categorizes rides into different seasons
based on the month in the “started_at” column. The following
seasons are used:
-
-
Creation:
- New clean and ready for analysis data frame “data_v2”.
- Exported CSV file “cyclistic_202208_202307.csv” from “data_V2”.
-
-
Guiding questions
- How should you organize your data to perform analysis on it?
- Has your data been properly formatted?
- What surprises did you discover in the data?
- What trends or relationships did you find in the data?
- How will these insights help answer your business questions?
-
Key tasks
- Aggregate your data so it’s useful and accessible.
- Organize and format your data.
- Perform calculations.
- Identify trends and relationships.
-
Deliverable
- A summary of your analysis
In the initial step of the analysis, a descriptive examination is performed for the two distinct rider types: casual and member.
The following statistical formulas and metrics are employed to describe the data frame:
-
Count (N): The count formula is used to determine the total number of observations within each rider type group.
-
Minimum (Min): By employing the minimum formula, the analysis identifies the smallest observed value for a particular variable.
-
Maximum (Max): The maximum formula is utilized to find the largest observed value, providing insights into the upper boundaries of a particular variable.
-
Mean (Average): The mean formula calculates the average value of the variable within each group. It offers an estimation of the central tendency or typical value for the variable in each group.
-
Median (Midpoint): The median formula computes the middle value when data is sorted. It serves as a measure of central tendency that is less influenced by extreme values, representing the “typical” value for the variable in each group.
-
Standard Deviation (SD): The standard deviation formula quantifies the dispersion of the variable within each group. A smaller standard deviation implies that data points are closer to the mean, whereas a larger standard deviation suggests greater variability.
summary_total <-
data_v2 %>%
group_by(member_casual) %>%
summarise(
count = n(),
min = hms::as_hms(min(ride_time_secs)),
max = hms::as_hms(max(ride_time_secs)),
mean = hms::as_hms(mean(ride_time_secs)),
median = hms::as_hms(median(ride_time_secs)),
sd = hms::as_hms(sd(ride_time_secs))
)
summary_total## # A tibble: 2 × 7
## member_casual count min max mean median sd
## <chr> <int> <time> <time> <time> <time> <time>
## 1 casual 1595997 00'01" 202:16:18 22'21.184391" 12'43" 47'50.006052"
## 2 member 2712633 00'01" 24:57:52 12'03.147822" 08'37" 20'08.588970"
The ride time average is influenced by outliers, as indicated by the maximum values and significant standard deviation. Therefore, in analyses where the data is affected by extreme values, it is advisable to use the median instead of the mean.
summary_period <- data_v2 %>%
mutate(period = factor(period, levels = c("Morning", "Lunch", "Afternoon", "Evening", "Night"))) %>%
group_by(member_casual, period) %>%
summarise(
count = n(),
median = hms::as_hms(median(ride_time_secs)))## `summarise()` has grouped output by 'member_casual'. You can override using the
## `.groups` argument.
summary_period %>% pivot_wider(names_from = member_casual, values_from = c(count, median))## # A tibble: 5 × 5
## period count_casual count_member median_casual median_member
## <fct> <int> <int> <time> <time>
## 1 Morning 245626 673591 10'11" 08'04"
## 2 Lunch 294646 421903 15'10" 08'12"
## 3 Afternoon 536353 869526 14'00" 09'06"
## 4 Evening 363253 588053 12'13" 08'59"
## 5 Night 156119 159560 10'43" 08'28"
Among the various periods of bike usage, the noticeable difference between users is in the ‘Morning’ period, which is predominantly active among members while not even top 3 for casual users. In terms of median ride durations, the ‘Lunch’ period boasts the longest rides for casual riders, while members have a more uniform length of rides across different periods. These statistics suggest that casual users may have a more tourist/leisure-oriented usage pattern, while members appear to have a usage pattern that aligns more with daily work routines. To confirm this pattern, it is essential to observe this behavior during different days, months, and seasons to gain a comprehensive understanding.
summary_weekday <- data_v2 %>%
mutate(day_of_week = factor(day_of_week, levels = c("Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"))) %>%
group_by(member_casual, day_of_week) %>%
summarise(
count = n(),
median = hms::as_hms(median(ride_time_secs)))## `summarise()` has grouped output by 'member_casual'. You can override using the
## `.groups` argument.
summary_weekday %>% pivot_wider(names_from = member_casual, values_from = c(count, median))## # A tibble: 7 × 5
## day_of_week count_casual count_member median_casual median_member
## <fct> <int> <int> <time> <time>
## 1 Monday 190106 390508 12'09" 08'13"
## 2 Tuesday 186853 429304 11'21" 08'22"
## 3 Wednesday 191358 436341 11'01" 08'26"
## 4 Thursday 210591 436569 11'19" 08'30"
## 5 Friday 242379 388869 12'35" 08'32"
## 6 Saturday 325803 340758 14'55" 09'33"
## 7 Sunday 248907 290284 14'43" 09'17"
summary_monthly <- data_v2 %>%
mutate(month = factor(month, levels = c("January", "February", "March", "April", "May", "June", "July", "August", "September", "October", "November", "December"))) %>%
group_by(member_casual, month) %>%
summarise(
count = n(),
median = hms::as_hms(median(ride_time_secs)))## `summarise()` has grouped output by 'member_casual'. You can override using the
## `.groups` argument.
summary_monthly %>% pivot_wider(names_from = member_casual, values_from = c(count, median))## # A tibble: 12 × 5
## month count_casual count_member median_casual median_member
## <fct> <int> <int> <time> <time>
## 1 January 29240 117796 08'14" 07'05.0"
## 2 February 32338 115968 09'16" 07'12.5"
## 3 March 46228 152641 09'13" 07'16.0"
## 4 April 109179 212020 12'16" 08'09.0"
## 5 May 175523 284336 13'57" 09'04.0"
## 6 June 218389 313158 13'47" 09'25.0"
## 7 July 243672 326883 14'21" 09'36.0"
## 8 August 268189 333125 13'44" 09'39.0"
## 9 September 219084 311913 12'50" 09'10.0"
## 10 October 150028 260782 11'44" 08'16.0"
## 11 November 72847 180749 09'59" 07'44.0"
## 12 December 31280 103262 08'38" 07'21.0"
summary_season <- data_v2 %>%
mutate(season = factor(season, levels = c("Winter", "Spring", "Summer", "Autumn"))) %>%
group_by(member_casual, season) %>%
summarise(
count = n(),
median = hms::as_hms(median(ride_time_secs)))## `summarise()` has grouped output by 'member_casual'. You can override using the
## `.groups` argument.
summary_season %>% pivot_wider(names_from = member_casual, values_from = c(count, median))## # A tibble: 4 × 5
## season count_casual count_member median_casual median_member
## <fct> <int> <int> <time> <time>
## 1 Winter 92858 337026 08'42" 07'13"
## 2 Spring 330930 648997 12'36" 08'18"
## 3 Summer 730250 973166 13'57" 09'34"
## 4 Autumn 441959 753444 11'56" 08'29"
It is now necessary to understand how users utilize each type of available ride differently.
summary_ride_type <- data_v2 %>%
mutate(rideable_type = factor(rideable_type, levels = c("classic_bike", "electric_bike", "docked_bike"))) %>%
group_by(member_casual, rideable_type) %>%
summarise(
count = n(),
median = hms::as_hms(median(ride_time_secs))) %>%
mutate(rideable_type = recode(rideable_type, "classic_bike" = "Classic", "electric_bike" = "Electric", "docked_bike" = "Docked"))## `summarise()` has grouped output by 'member_casual'. You can override using the
## `.groups` argument.
summary_ride_type %>% pivot_wider(names_from = member_casual, values_from = c(count, median))## # A tibble: 3 × 5
## rideable_type count_casual count_member median_casual median_member
## <fct> <int> <int> <time> <time>
## 1 Classic 782927 1680623 14'00" 09'03"
## 2 Electric 688107 1032010 10'16" 08'02"
## 3 Docked 124963 NA 26'47" NA
In the table provided above, a comprehensive summary of ride types, including ‘Classic,’ ‘Electric,’ and ‘Docked,’ is presented. The data reveals that riders exhibit a preference for ‘Classic’ and ‘Electric’ types, with ‘Classic’ being the most favored option among them. Notably, ‘Docked’ types are only used by casual riders and appear to be less frequently utilized, however, they have the longest duration of usage.
Finally, which stations are most used by users.
top10station_casual <- data_v2 %>%
filter(member_casual == "casual") %>%
count(start_station_name) %>%
arrange(desc(n)) %>%
head(10)
top10station_casual## start_station_name n
## 1: Streeter Dr & Grand Ave 47048
## 2: DuSable Lake Shore Dr & Monroe St 28465
## 3: Michigan Ave & Oak St 21493
## 4: Millennium Park 20902
## 5: DuSable Lake Shore Dr & North Blvd 19342
## 6: Shedd Aquarium 17791
## 7: Theater on the Lake 15421
## 8: Dusable Harbor 13004
## 9: Wells St & Concord Ln 12666
## 10: Indiana Ave & Roosevelt Rd 11482
top10station_member <- data_v2 %>%
filter(member_casual == "member") %>%
count(start_station_name) %>%
arrange(desc(n)) %>%
head(10)
top10station_member## start_station_name n
## 1: Kingsbury St & Kinzie St 23417
## 2: Clark St & Elm St 22140
## 3: Clinton St & Washington Blvd 21573
## 4: Wells St & Concord Ln 19418
## 5: Loomis St & Lexington St 19389
## 6: University Ave & 57th St 18595
## 7: Ellis Ave & 60th St 18222
## 8: Clinton St & Madison St 17892
## 9: Wells St & Elm St 17640
## 10: Canal St & Adams St 16647
The top 10 stations by users shows that they use different stations. Casual riders, in particular, utilize the station at Streeter Dr & Grand Ave nearly twice as often as the second most-used station by them.
-
Ride Frequency:
- Members tend to have a higher frequency of rides compared to casual riders.
-
Ride Duration:
- Casual riders generally have longer ride durations than members.
-
Usage Patterns by Time of Day:
- The noticeable difference between users is in the ‘Morning’ period, which is predominantly active among members while not even top 3 for casual users. Median ride durations for casual riders are longest during the ‘Lunch’ period, while members exhibit more consistent ride durations across different time periods.
-
Weekday vs. Weekend Usage:
- Casual riders have a higher ride frequency on weekends, while members ride more frequently on weekdays. Interestingly, both casual riders and members have longer ride durations on weekends, which suggests that weekends might be a preferred time for more leisure bike trips for both groups.
-
Seasonal Trends:
- Summer and autumn are the most popular seasons for bike rides, with casual riders having a significantly higher number of rides in the summer compared to other seasons. This could indicate that casual riders are more inclined to use bikes during vacation seasons.
-
Ride Types:
- ‘Classic’ and ‘Electric’ ride types are favored by riders, with ‘Classic’ being the most popular choice. ‘Docked’ ride types are only used by casual riders and have the longest ride durations. In contrast, members mainly opt for ‘Classic’ rides, which tend to have shorter durations. This suggests that members may prefer shorter, more efficient rides for their daily routines, while casual riders may use ‘Docked’ bikes for longer, leisurely trips.
-
Top 10 Stations:
- It’s worth noting that the top 10 stations used by members and casual riders are different, indicating distinct preferences or usage patterns for station selection among these two groups.
-
Guiding questions
- Were you able to answer the question of how annual members and casual riders use Cyclistic bikes differently?
- What story does your data tell?
- How do your findings relate to your original question?
- Who is your audience? What is the best way to communicate with them?
- Can data visualization help you share your findings?
- Is your presentation accessible to your audience?
-
Key tasks
- Determine the best way to share your findings.
- Create effective data visualizations.
- Present your findings.
- Ensure your work is accessible.
-
Deliverable
- Supporting visualizations and key findings
The analysis will be shared through R charts in this notebook, and a Slide presentation.
The first plot displays a daily breakdown of the number of rides per user, illustrating that only a few days saw casual users utilize the bike share system more frequently than members. Seasonal patterns reveal distinct usage trends, with summer showing the highest activity for members averaging just over 10,000 daily rides and casual users around 7,500 daily rides. During winter, member usage remains relatively steady, slightly above 3,000 daily rides, while casual users decrease to around 1,000 daily rides.
summary_date <- data_v2 %>%
group_by(member_casual, date) %>%
summarise(count = n())## `summarise()` has grouped output by 'member_casual'. You can override using the
## `.groups` argument.
summer_period <- data.frame(
start_date = as.Date(c("2022-08-01", "2023-06-01")),
mid_date = as.Date(c("2022-08-15", "2023-07-01")),
end_date = as.Date(c("2022-08-31", "2023-07-31"))
)
num_rides_day <- ggplot(summary_date) +
geom_area(aes(x = date, y = count, fill = member_casual), position = "identity", alpha = 0.6) +
geom_smooth(aes(x = date, y = count, color = member_casual), method = "loess", span = 0.1, se = FALSE, size = 1.2) +
geom_rect(data = summer_period, aes(xmin = start_date, xmax = end_date, ymin = -Inf, ymax = Inf), alpha = 0.1) +
geom_text(data = summer_period, aes(x = mid_date, y = 0, label = "Summer"), vjust = -1) +
scale_fill_viridis_d() +
scale_color_viridis_d() +
labs(title = "Total Number of Rides per Day by User", x = NULL, y = NULL, fill = NULL, color = NULL, caption = "August 2022 - July 2023") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5))## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
## ℹ Please use `linewidth` instead.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.
ggsave(filename = "num_rides_day.png", plot = num_rides_day)## Saving 7 x 5 in image
## `geom_smooth()` using formula = 'y ~ x'
num_rides_day## `geom_smooth()` using formula = 'y ~ x'
Looking at the median duration of rides per month, casual riders exhibit a higher value with significant variance during hot seasons, while members maintain a steady usage pattern, similar to routine patterns.
dur_rides_month <- ggplot(summary_monthly) +
geom_line(aes(x = month, y = median, color = member_casual, group = member_casual, linetype = member_casual), size = 1.5) +
geom_rect(aes(xmin = "March", xmax = "November", ymin = -Inf, ymax = Inf), alpha = 0.01) +
geom_text(x = "July", y = 0, label = "Duration of casual users peaks during 'hot' months", hjust = 0.5, vjust = -1) +
geom_text(data = summary_monthly %>% filter(month == "January", member_casual == "casual"), aes(x = "January", y = median, label = median), hjust = 0.5, vjust = -2, nudge_x = 0.1) +
geom_text(data = summary_monthly %>% filter(month == "July", member_casual == "casual"), aes(x = "July", y = median, label = median), hjust = 0.5, vjust = -2, nudge_x = 0.1) +
scale_color_viridis_d() +
labs(title = "Median Duration of Rides per Month by User", x = NULL, y = NULL, linetype = NULL, color = NULL, caption = "August 2022 - July 2023") +
scale_linetype_manual(values = c("longdash", "solid")) +
scale_y_time(limits = c(hms::as_hms(0), hms::as_hms(20*60))) +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1), plot.title = element_text(hjust = 0.5))
ggsave(filename = "dur_rides_month.png", plot = dur_rides_month)## Saving 7 x 5 in image
dur_rides_month
The number of rides per weekday demonstrates an inverse pattern between users. Casual riders tend to use the service more frequently on weekends, while members show higher usage during the week. This difference indicates distinct usage patterns, possibly related to tourist and work routines. The peak usage times also vary, with casual riders favoring the afternoon and evening, while members predominantly ride in the morning and afternoon.
summary_period_of_day <- data_v2 %>%
mutate(day_of_week = factor(day_of_week, levels = c("Sunday", "Saturday", "Friday", "Thursday", "Wednesday", "Tuesday", "Monday"))) %>%
mutate(period = factor(period, levels = c("Night", "Evening", "Afternoon", "Lunch", "Morning"))) %>%
group_by(member_casual, day_of_week, period) %>%
summarise(count = n())## `summarise()` has grouped output by 'member_casual', 'day_of_week'. You can
## override using the `.groups` argument.
num_rides_weekday_period <- ggplot(summary_period_of_day) +
geom_col(aes(x = day_of_week, y = count, fill = period), position = "stack") +
scale_fill_viridis_d() +
labs(title = "Total Number of Rides per Weekday and Period of Day by User", x = NULL, y = NULL, fill = NULL, caption = "August 2022 - July 2023") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, hjust = 1), plot.title = element_text(hjust = 0.5)) +
facet_wrap(~member_casual, labeller = labeller(member_casual = c("casual" = "Casual", "member" = "Member"))) +
coord_flip()
ggsave(filename = "num_rides_weekday_period.png", plot = num_rides_weekday_period)## Saving 7 x 5 in image
num_rides_weekday_period
There are three types of rides available: Classic, Electric, and Docked. Docked rides are exclusively used by casual riders, accounting for only 8% of their choices. Classic rides are the preferred option for both user groups, with members choosing Classic 62% of the time.
summary_ride_type <- summary_ride_type %>%
mutate(percentage = count / sum(count)) %>%
arrange(member_casual, percentage)
num_ridetype <- ggplot(summary_ride_type) +
geom_bar(aes(x = "", y = percentage, fill = rideable_type), stat = "identity", color = "black") +
geom_text(aes(x = 1.6, y = percentage, label = paste0(round(percentage*100), "%")), color = "black", position = position_stack(vjust = 0.5)) +
coord_polar(theta = "y") +
scale_fill_viridis_d() +
labs(title = "Ride Type Distribution by User", fill = NULL, caption = "August 2022 - July 2023") +
theme_void() +
theme(plot.title = element_text(hjust = 0.5)) +
facet_wrap(~ member_casual, labeller = labeller(member_casual = c("casual" = "Casual", "member" = "Member")))
ggsave(filename = "num_ridetype.png", plot = num_ridetype)## Saving 7 x 5 in image
num_ridetype
Concerning ride duration, classic rides are also favored over electric rides by both user groups. However, docked rides are used for twice the duration that casual riders spend on classic rides. While members maintain a relatively consistent ride duration across seasons, casual users have significantly shorter rides during winter.
summary_ride_type_season <- data_v2 %>%
mutate(rideable_type = factor(rideable_type, levels = c("classic_bike", "electric_bike", "docked_bike"))) %>%
group_by(member_casual, season, rideable_type) %>%
summarise(median = hms::as_hms(median(ride_time_secs))) %>%
mutate(rideable_type = recode(rideable_type, "classic_bike" = "Classic", "electric_bike" = "Electric", "docked_bike" = "Docked"))## `summarise()` has grouped output by 'member_casual', 'season'. You can override
## using the `.groups` argument.
dur_ride_type <- ggplot(summary_ride_type_season) +
geom_col(aes(x = season, y = median, fill = rideable_type), position = "dodge") +
geom_text(data = summary_ride_type_season %>% filter(rideable_type %in% c("Classic", "Docked"), member_casual == "casual"), aes(x = season, y = median, label = substr(median, 4, 8)), vjust = -0.5) +
scale_fill_viridis_d() +
scale_y_time(limits = c(hms::as_hms(0), hms::as_hms(30*60))) +
labs(title = "Median Duration of Rides per Season and Ride Type by User", fill = NULL, x = NULL, y = NULL, caption = "August 2022 - July 2023") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5)) +
facet_wrap(~member_casual, labeller = labeller(member_casual = c("casual" = "Casual", "member" = "Member")))
ggsave(filename = "dur_ride_type.png", plot = dur_ride_type)## Saving 7 x 5 in image
dur_ride_type
Upon examining the top 10 stations where users commence their rides, it becomes evident that a tourist pattern prevails among casual riders, as they tend to favor stations situated closer to the coastline of Chicago. Conversely, members exhibit a preference for inner-city stations, aligning with a work-related behavior. Additionally, there is one station (Streeter Dr & Grand Ave) that stands out with double the usage compared to others among casual users, while members distribute their usage more evenly across all stations.
top10station_casual_density <- top10station_casual %>% left_join(data_v2 %>% filter(member_casual == "casual"), by = "start_station_name")
top10station_member_density <- top10station_member %>% left_join(data_v2 %>% filter(member_casual == "member"), by = "start_station_name")
map <- get_stamenmap(bbox = c(
min(top10station_member_density$start_lng),
min(top10station_member_density$start_lat),
max(top10station_member_density$start_lng),
max(top10station_member_density$start_lat)
), maptype = "terrain")## ℹ Map tiles by Stamen Design, under CC BY 3.0. Data by OpenStreetMap, under ODbL.
top10stations <- bind_rows(top10station_casual_density %>% distinct(start_station_name, .keep_all = TRUE), top10station_member_density %>% distinct(start_station_name, .keep_all = TRUE))
top_station_locations <- ggmap(map) +
geom_point(data = top10stations, aes(x = start_lng, y = start_lat, color = member_casual, size = n), alpha = 0.8) +
scale_size(range = c(.1, 20), name="Density") +
scale_color_viridis_d() +
labs(title = "Top 10 Start Stations Locations by User", color = "User Type", x = NULL, y = NULL, caption = "August 2022 - July 2023") +
theme_minimal() +
theme(plot.title = element_text(hjust = 0.5))
ggsave(filename = "top_station_locations.png", plot = top_station_locations)## Saving 7 x 5 in image
top_station_locations
-
Guiding questions
- What is your final conclusion based on your analysis?
- How could your team and business apply your insights?
- What next steps would you or your stakeholders take based on your findings?
- Is there additional data you could use to expand on your findings?
-
Key tasks
- Create your portfolio.
- Add your case study.
- Practice presenting your case study to a friend or family member.
-
Deliverable
- Your top three recommendations based on your analysis
In the analysis of the bike share company’s user behavior, significant insights have emerged regarding casual users. These individuals tend to exhibit leisure and tourist-oriented behaviors, utilizing the bike system for extended durations, primarily during weekends and the hot seasons, notably in the summer. Of particular importance, casual users are the only users of docked bikes, even though this ride type constitutes only 8% of their total rides, it delivers the longest durations of rides, twice compared to other types. Additionally, casual users display a preference for stations located near Chicago’s picturesque coastline. Armed with these insights, it was developed a tailored marketing campaign with three strategic recommendations to convert these casual users into annual members.
1. Seasonal Membership Promotions:
- Target casual users' preference for leisure and tourist behavior during the hot season, especially in summer. Create special seasonal promotions that make annual memberships more appealing during this time. For instance, Cyclistic could offer discounted annual memberships during the summer months, along with additional perks such as free helmets, guided tours, or access to exclusive events. Highlight the benefits of an annual membership, such as unlimited access and convenience, which can enhance their leisurely bike rides along the coastline.
2. Docked Bike Experience Enhancement:
- Since casual users show a preference for docked bikes using this type longer than others, focus on improving their experience with this type. Highlight the longer ride durations and convenience of docked bikes compared to classic and electric ones. Consider launching marketing campaigns that educate casual users about the advantages of docked bikes, such as stability, comfort, and their suitability for leisurely rides. Offer special promotions or discounts for annual memberships, with a focus on the use of docked bikes, and create digital media content that showcases scenic routes along the coastline that are best enjoyed with docked bikes.
3. Weekend Getaway Packages:
- Recognizing that casual users often ride on weekends, design packages that encourage them to become annual members. Develop special weekend getaway packages that include an annual membership, a list of popular weekend destinations along the coastline, and discounts at partner businesses (e.g., restaurants, ice cream shops, or museums). The social media marketing campaign should use the area range of preferred stations near the coastline and commence around lunchtime, because casual users increase their activity after morning period. The campaign's focal point should be effectively showcase the exceptional experiences offered by the annual membership through these weekend packages.