Cyclistic_Case_Study_for_Data_Analysis

Title: Cyclistics Data Analysis

Author: Kareem Mohamed

Date: 15-3-2024

Reference

Scenario

The director of marketing believes the company’s future success depends on maximizing the number of annual memberships. Therefore, the team wants to understand how casual riders and annual members use Cyclistic bikes differently. From these insights, the marketing analyst team at Cyclistic that I'm a part of will design a new marketing strategy to convert casual riders into annual members. Moreno believes that maximizing the number of annual members will be key to future growth. Cyclistic’s finance analysts have concluded that annual members are much more profitable than casual riders.

About The Company

Cyclistic, a bike-share company in Chicago, in 2016, Cyclistic launched a successful bike-share offering. Since then, the program has grown to a fleet of 5,824 bicycles that are geotracked and locked into a network of 692 stations across Chicago. The bikes can be unlocked from one station and returned to any other station in the system at a time.

Cyclistic’s marketing strategy relied on building general awareness and appealing to broad consumer segments. One approach that helped make these things possible was the flexibility of its pricing plans: single-ride passes, full-day passes, and annual memberships.

Ask

Business Task

Design marketing strategies aimed at converting casual riders into annual members by analyzing the historical bike trip data given by Cyclistic to identify trends.

Key Stakeholder

Lily Moreno <- Director of marketing and my manager.
Cyclistic Marketing 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.

Prepare

Cyclistic’s historical trip data is public data that I used to analyze how different customer behaviors are using Cyclistic bikes, the data is located here. It is stored in csv file format and it contains 12 files for the past 12 months starting from February 2024 and going back to March 2023.

The data has some limitations :

It has a lot of empty cells in the start_station and end_station as well as the station_id.

Data privacy issues prohibit me from using riders’ personally identifiable information. This means that I won’t be able to connect pass purchases to credit card numbers to determine if casual riders live in the Cyclistic service area or if they have purchased multiple single passes.

I'm going to check the data using the ROCCC approach

Reliable: The data is reliable as it comes from Bike Share which operates the City of Chicago, the city permits Bike Share to make certain Divvy system data owned by the City available to the public.
Original: The data was gathered and released to the public by Motivate Internation Inc., which makes the data original.
Comprehensive: The data is organized in a long format and contains various columns such as rider_id, rideable_type, started_at, ended_at, start_staion_name, end_station_name, start_station_id, end_station_id, member_causal.
Current: The data is up to date as it has data for the past 12 months till February.
Cited: Bikeshare hereby grants a non-exclusive, royalty-free, limited, perpetual license to access, reproduce, analyze, copy, modify, distribute in your product or service and use the Data for any lawful purpose License.

Process

What tools did I use ? and Why?

I used Excel and R to check, clean, and manipulate the data but the analyzing was done on R. The data contains a lot of information that R would handle better than Excel so that's why I used R in manipulating and cleaning most of the time.

Documentation of data cleaning in Excel :

I checked the data for duplicates but none of them had any duplicates.
I changed each file name by naming it based on the month and year that it contains data for, so for example "202402-divvy-tripdata" is now called "february24-divvy-tripdata".

Documentation of data cleaning in R :

-Importing Data

Importing the data that I downloaded and combining it in one data frame called combined_bike_data, the data frame will combine 12 CSV files containing historical data about the past 12 months of Chicago city bike riders starting from February 2024 going back to March 2023.

knitr::opts_knit$set(root.dir = "D:/Coursera Data Analysis/Cyclistic")

After combining the data the total observations are {5,707,168} and it has {15} variables.

combined_bike_data <- rbind(
read.csv("february24-divvy-tripdata.csv"),
read.csv("january24-divvy-tripdata.csv"),
read.csv("december23-divvy-tripdata.csv"),
read.csv("november23-divvy-tripdata.csv"),
read.csv("october23-divvy-tripdata.csv"),
read.csv("september23-divvy-tripdata.csv"),
read.csv("august23-divvy-tripdata.csv"),
read.csv("july23-divvy-tripdata.csv"),
read.csv("june23-divvy-tripdata.csv"),
read.csv("may23-divvy-tripdata.csv"),
read.csv("april23-divvy-tripdata.csv"),
read.csv("march23-divvy-tripdata.csv")
)

Inspecting

I start by inspecting various data frames using glimpse(), str(), and colnames().

glimpse(combined_bike_data)
colnames(combined_bike_data)

Adding N/A to empty cells

Some cells in the data frame are empty they don't have N/A values so in order to delete them or to acknowledge them I will add NA to them.

combined_bike_data$start_station_name[combined_bike_data$start_station_name == ""] <- NA
combined_bike_data$start_station_id[combined_bike_data$start_station_id == ""] <- NA
combined_bike_data$end_station_name[combined_bike_data$end_station_name == ""] <- NA
combined_bike_data$end_station_id[combined_bike_data$end_station_id == ""] <- NA

Adding columns

1- Date, Month, Day, and year

Adding columns for date, month, day, and year by using as.Date() on started_at to get the date and then getting the rest of the new columns.

combined_bike_data$date <- as.Date(combined_bike_data$started_at)
combined_bike_data$month <-  format(as.Date(combined_bike_data$date), "%m")
combined_bike_data$day <- format(as.Date(combined_bike_data$date), "%d")
combined_bike_data$year <- format(as.Date(combined_bike_data$date), "%Y")

2- Ride Length

First, I'm going to identify the correct format of started_at and ended_at

unique(combined_bike_data$started_at)

As you can see the format is YYY-MM-DD HH:MM:SS. changing the data type in started_at and ended_at to the right format.

combined_bike_data$started_at <- as.POSIXct(combined_bike_data$started_at, format = "%Y-%m-%d %H:%M:%S")
combined_bike_data$ended_at <- as.POSIXct(combined_bike_data$ended_at, format = "%Y-%m-%d %H:%M:%S")

Adding a column for ride length by deducting ended_at from started_at

combined_bike_data <- combined_bike_data %>% mutate(ride_length = ended_at - started_at)

dividing the ride_length by 60 to represent it by minutes not seconds.

combined_bike_data$ride_length <- as.numeric(combined_bike_data$ride_length)
combined_bike_data$ride_length <- as.numeric(combined_bike_data$ride_length/60)

3- Days of The Week

Adding column day_of_week to know the days of the week that the rides started_at.

combined_bike_data <- combined_bike_data %>%
  mutate(day_of_week = weekdays(as.Date(combined_bike_data$started_at)))

Removing Columns

Removing data that are not useful for the analysis, so I removed start_lat, start_lng, end_lat, and end_lng.

combined_bike_data <- combined_bike_data %>%
  select(-c(start_lat, start_lng, end_lat, end_lng))

Adding N/A to specific cells

Some cells need to be removed so I will add NA to them based on some specific conditions.

First ride_length has some values that are equal to 0 and negative number that cannot be displayed which means that the rider hasn't used the bike for any hours, minutes, or seconds or the ride length is negative as well, so I will not be using these cells in my analysis.

Then using sum(is.na) to know how many ride_length were equal to 0 and negative numbers, as you can see we had {1560} entry.

combined_bike_data$ride_length[combined_bike_data$ride_length <= "0"] <- NA
sum(is.na(combined_bike_data$ride_length))

Then I will subset the data to make further adjustments to the data

subset_data <- combined_bike_data[!is.na(combined_bike_data$ride_length), ]

Next adding NA to values that are more than one day because we have some values in ride_length that are more than 12 thousand minutes which is not normal so any values that is more than one day will be replaced by NA.

ride_length that is more than 1440 minutes = one day.

Using sum(is.na) I found that the values that are more than 1440 minutes are 6882 entries.

subset_data[subset_data$ride_length > 1440, "ride_length"] <- NA
sum(is.na(subset_data$ride_length))

combining the subsets back to the original data. and using sum(is.na) the total N/A's in the ride_length of the combined_bike_data is 8442 entry.

combined_bike_data[!is.na(combined_bike_data$ride_length), ] <- subset_data
sum(is.na(combined_bike_data$ride_length))

Dropping the subset data.

rm(subset_data)

Deleting N/A Values

Using na.omit() to delete the N/A values in the data frame.

combined_bike_data <- na.omit(combined_bike_data)

There were {1,375,436} observations deleted and now the available observations that I will use for analysis are {4,331,652} observations, mostly the deleted N/A's were from the start_station_name, end_station_name, start_station_id, end_station_id and some rows in the ride_length.

Analyze

Ride Length

Calculating the mean, min, and max for the ride length

mean is {15.9} minutes , min is {0.016} and max is {1439.8} minutes.

mean(combined_bike_data$ride_length) 
min(combined_bike_data$ride_length)
max(combined_bike_data$ride_length)

Ride Length and Rideable Type

Doing a summary to see what type of ride is being used more when the ride length is more than one hour and less than one hour.

The first condition is using the electric and classic bikes when the length is more than 60 minutes.

the outcome is that {298,193} rides were electric bikes and {494,360} rides were classic bikes.

The second condition is using the same rideable types when the ride length is shorter than 60 minutes. the outcome is that {1,234,413} rides were by electric bikes and {2,232,262} were by classic bikes. The conclusion is that classic bikes are being used more when the ride length is less or more than 60 minutes, but more riders ride the bike for a preiod shorter than 60 minutes.

sum(combined_bike_data$rideable_type == "electric_bike" & combined_bike_data$ride_length > "60")
sum(combined_bike_data$rideable_type == "classic_bike" & combined_bike_data$ride_length > "60")
sum(combined_bike_data$rideable_type == "electric_bike" & combined_bike_data$ride_length < "60")
sum(combined_bike_data$rideable_type == "classic_bike" & combined_bike_data$ride_length < "60")

By using the sum you can see that the total electric bike used from the total rides, {1,532,620} rides were by electric bikes and {2,726,663} rides were by classic bike so the classic bikes are being used more.

sum(combined_bike_data$rideable_type == "electric_bike")
sum(combined_bike_data$rideable_type == "classic_bike")

Member and Casual Riders Count

Using sum to know how many members and casual riders Cyclistics had throughout the past 12 months. they had {2,806,688} members and {1,524,848} casual riders.

sum(combined_bike_data$member_casual == "member")
sum(combined_bike_data$member_casual == "casual")

Member and Casual Riders / Electric and Classic bikes

Using sum to see who uses electric bikes and classic bikes more, is it members or is it casual riders?

First, members use more electric bikes than the casual riders as {961, 489} rides were by members and {571,131} rides were by casual riders.

Second, members also use more classic bikes than casual riders as {1,845,199} rides were by members and {881,464} rides were by casual riders.

Casual riders and members tend to use classic bikes more than electric bikes.

sum(combined_bike_data$rideable_type == "electric_bike" & combined_bike_data$member_casual == "member")
sum(combined_bike_data$rideable_type == "electric_bike" & combined_bike_data$member_casual == "casual")
sum(combined_bike_data$rideable_type == "classic_bike" & combined_bike_data$member_casual == "member")
sum(combined_bike_data$rideable_type == "classic_bike" & combined_bike_data$member_casual == "casual")

Riders by Day of Week

Using the combined_bike_data data frame we can see that members ride bikes the most on Thursdays and casual riders ride bikes more on Saturdays, this shows that there is a difference in the riding behavior between members and casual riders

member_casual_color <- c (
"casual" = "brown",
"member" = "black"
)
combined_bike_data %>%
  group_by(member_casual, day_of_week) %>%
  summarise(count = n()) %>%
  ggplot(aes(x = day_of_week, y = count, fill = member_casual)) +
  geom_col(position = "dodge") +
  geom_text(aes(label = round(count, 2)), position = position_dodge(width = 1.1), vjust = -0.5, size = 3) +
  scale_fill_manual(values = member_casual_color, c("Member_Casual"))+
  labs(x = "Day Of Week", y = "Member-Casual Count", title = "Member vs Casual Riders By Week Days")+
  theme_minimal()

Riders by Month

Using the same inputs from the last visualization but changing the group_by to group it by month.

The visualization shows that for members August has the highest number of member riders for the past 12 months and Casual riders are riding bikes the most in July, overall June, July, August, and September are the highest 4 months of the past 12 that have the highest riders members and casual riders.

The lowest months with riders are December, January, February, and March, they have the lowest number of riders for the past 12 months, this means that riders either members or casual riders usually ride the most in summer and ride fewer in the winter.

combined_bike_data %>%
  group_by(month, member_casual) %>%
  summarise(count = n()) %>%
  ggplot(aes(x = month, y = count, fill = member_casual)) +
  geom_col(position = "dodge") +
  geom_text(aes(label = round(count, 2)), position = position_dodge(width = 0.7 ), vjust =    -0.5 ,size = 3, color = "darkgrey") +
  scale_fill_manual(values = member_casual_color,c("Member_Casual"))+
  labs(x = "Month", y = "Member-Casual Count", title = "Member Vs Casual Rides By Month")+
  theme_minimal()

Classic Bike vs Electric Bike

This visualization shows a comparison between classic bikes and electric bikes based on usage by casual riders and members.

Classic bikes are being used more by both members and casual riders but the overall usage of bikes is more by members.

combined_bike_data %>%
  filter(rideable_type %in% c("classic_bike", "electric_bike")) %>%
  group_by(rideable_type, member_casual) %>%
  summarise(count = n()) %>%
  ggplot(aes(x = rideable_type, y = count, fill = member_casual))+
  geom_col(position = "dodge") +
  geom_text(aes(label = round(count, 2)), position = position_dodge(width = 0.7 ), vjust = -0.5 ,size = 3) +
  scale_fill_manual(values = member_casual_color, c("Member_Casual")) + 
  labs(x = "Rideable Type", y = "Member-Casual Count", title = "Classic Bike Vs Electric Bike") +
  theme_minimal()

AVG Ride Length

comparing members vs casual riders based ob=n their ride length to see what are their riding habits.

This visualization shows that casual riders tend to spend more time riding as their average rides exceed 22.92 minutes, and members spend much less time as their average rides length is shorter than 12.23 minutes.

combined_bike_data %>%
  group_by(member_casual) %>%
  summarise(avg_ride_length = mean(ride_length, na.rm = TRUE)) %>%
  ggplot(aes(x = member_casual, y = avg_ride_length, fill = member_casual)) +
  geom_col(position = "dodge") +
  geom_text(aes(label = round(avg_ride_length, 2)), position = position_dodge(width = 0.9), vjust = -0.5, size = 3) +
  scale_fill_manual(values = member_casual_color, c("Member_Casual"))+
  labs(x = "Member_Casual", y = "Ride Length", title = "Member vs Casual Riders By Ride Length")+
  theme_minimal()

AVG Ride Length by Weekday

Comparing the avg ride_length by weekdays and rideable_type. for casual riders the longest they ride is on Sundays with more than an average of 26.8 minutes and members as well are on Sundays but with an average of almost 13.71 minutes. the shortest they ride are on Wednesdays for casual riders with an average of 19.62 minutes and Mondays and Thursdays for members with an average of 11.68 minutes.

combined_bike_data %>%
  group_by(day_of_week, member_casual) %>%
  summarise(avg_ride_length = mean(ride_length, na.rm = TRUE)) %>%
  ggplot(aes(x = day_of_week, y = avg_ride_length, fill = member_casual)) +
  geom_col(position = "dodge") +
  geom_text(aes(label = round(avg_ride_length, 2)), position = position_dodge(width = 0.9), vjust = -0.5, size = 3) +
  scale_fill_manual(values = member_casual_color,c("Member_Casual"))+
  labs(x = "Day Of Week", y = "Average Ride Length", title = "Average Ride Length by Weekday and Member_Casual")+
  theme_minimal()

AVG Ride Length and Rideable Type

Comparing the avg ride length based on the rideable type. The visualization shows that the classic bike has an average length of 17.17 minutes and the electric bike has an average of 12.14 minutes, which means that the classic bike is used more for long trips and the electric bike is for shorter rides.

classic_electric <- c (
  "classic_bike" = "brown",
  "electric_bike" = "black"
)
combined_bike_data %>%
  filter(rideable_type %in% c("classic_bike", "electric_bike")) %>%
  group_by(rideable_type) %>%
  summarise(avg_ride_length = mean(ride_length, na.rm = TRUE)) %>%
  ggplot(aes(x = rideable_type, y = avg_ride_length, fill = rideable_type)) +
  geom_col(position = "dodge") +
  geom_text(aes(label = round(avg_ride_length, 2)), position = position_dodge(width = 0.9), vjust = -0.5, size = 3) +
  scale_fill_manual(values = classic_electric)+
  labs(x = "Rideable Type", y = "Average Ride Length", title = "Average Ride Length by Rideable Type")+
  theme_minimal()

Dashboard

Tableau Cyclistic Dashboard

Act

Statistical Summary

Classic bikes are used more when the ride length is longer or shorter than 60 minutes as it is more reliable than electric bikes either for casual riders or member.
6.8% of riders who ride for more than one hour use electric bikes and 11.4% use classic bikes.
28.4% of riders that ride for less than one hour use electric bikes and 51.5% use classic bikes.
64.7% of riders are members and 35.3% are casual riders.
Overall 35.3% of rides were on electric bikes 62.9% were on classic bikes and the rest were docked bikes.
22.19% of rides by members were on electric bikes 42.5% were on classic bikes, 13.18% of rides by casual riders were on electric bikes and 20.3% were on classic bikes and 1.67% were docked bikes.
Classic bikes are used more than electric bikes by casual riders and members.
Members increase riding bikes on Thursdays and casual riders increase riding on Saturdays.
June, July, August, and September have the highest number of riders with more than 300,000 riders going up to 350,000 riders, December, Jan, and February have the lowest number of riders which means that in summer there are more riders than in winter.
The average ride length for members is 12.23 minutes and for casual riders is 22.92 minutes.
For casual riders, the longest they ride is on Sundays with an average of 26.8 minutes, and for members, it is as well on Sundays but with an average of 13.71 minutes.
The shortest they ride are on Wednesdays for casual riders with an average of 19.62 minutes, and Mondays and Thursdays for members with an average of 11.68 minutes.
The average ride length done using a classic bike is 17.17 minutes and using electric bikes is 12.14 minutes.

Recommendations

Target casual riders in the summer because they used to ride the most in summer so try and make them specific discounts or targeted ads in these 4 months.
Casual riders ride more on Saturdays, send them notifications or ads on weekends to make them aware of the membership subscription.
Casual riders ride for a longer time on Sundays so we can give them free rides on some specific weekends so to build customer connection with the casual riders.
Give casual riders a mini subscription to let them try how it feels to own a longer membership rather than the daily pay.

karemelshimi / cyclistic_case_study_for_data_analysis Goto Github PK