Source Code
All source code is written in the R Programming Language (R Core Team 2019) using the R Studio IDE (RStudio Team 2020). The packages used include
the popular tidyverse (Wickham et al.
2019) set of packages for doing most of the heavy lifting in the
code, knitr (Xie 2015) for rendering the R
Markdown (Allaire et al. 2023) documents,
and skimr (Waring et al. 2022) for
creating detailed summaries of the data. Finally, for generating the
visuals, we used the ggplot2 (Wickham
2016) package.
# set a seed in case we use any random numbers
set.seed(1337)
# Set the names of the packages and libraries you want to install
required_libraries <- c("tidyverse", "purrr", "knitr", "skimr", "readr")
# Install missing packages and load all required libraries
for (lib in required_libraries) {
if (!requireNamespace(lib, quietly = TRUE)) {
install.packages(lib)
}
library(lib, character.only = TRUE)
}
# load all the cycling data for analysis
# our data runs from 04/2022 to 04/2023
if(file.exists("combined_data_df.rds")) {
# read in our combined data
combined_data_df <- readRDS("combined_data_df.rds")
} else {
# get file list
file_list <- list.files(path = "SourceData/", pattern = "*.csv", full.names = TRUE)
# read and combine files
combined_data_df <- map_df(file_list, read_csv)
# convert ridable_type and member_casual into factors
combined_data_df$rideable_type <- as.factor(combined_data_df$rideable_type)
combined_data_df$member_casual <- as.factor(combined_data_df$member_casual)
# save the combined_data object so we don't have to reload this data again
# unless it changes
write_rds(combined_data_df, "combined_data_df.rds")
}
# examine the structure of the data
str(combined_data_df)
# Identify duplicates
duplicates <- duplicated(combined_data_df)
# Count duplicates
num_duplicates <- sum(duplicates)
# Show the number of duplicates
cat("Total number of duplicate rows found:",num_duplicates)
# Check for missing values
colSums(is.na(combined_data_df))
# Print detailed summary statistics of the dataframe
skim(combined_data_df)
# calculate the total ride time in minutes for each group
combined_data_df$ride_length <-
round((combined_data_df$ended_at - combined_data_df$started_at) / 60, 2)
# Change the units attribute to "minutes"
attr(combined_data_df$ride_length, "units") <- "minutes"
# Add a new column for the date without the time
combined_data_df$started_date <- as.Date(combined_data_df$started_at)
# Create the 'day_of_week_start' and 'day_of_week_end' columns
combined_data_df$day_of_week_start <- wday(combined_data_df$started_at, label = TRUE)
combined_data_df$day_of_week_end <- wday(combined_data_df$ended_at, label = TRUE)
# Convert the new columns to factors
combined_data_df$day_of_week_start <- as.factor(combined_data_df$day_of_week_start)
combined_data_df$day_of_week_end <- as.factor(combined_data_df$day_of_week_end)
# Get rid of outliers by only accepting ride times that are less than two days
# and greater than zero
combined_data_df <- combined_data_df %>% filter(ride_length <= 2880 &
ride_length > 5)
# split our data into two groups: member and casual
member_df <- combined_data_df %>% filter(member_casual == "member")
casual_df <- combined_data_df %>% filter(member_casual == "casual")
# ride_length metrics for both groups
ride_length_mean_member <- mean(as.numeric(member_df$ride_length))
ride_length_max_member <- max(as.numeric(member_df$ride_length))
ride_length_min_member <- min(as.numeric(member_df$ride_length))
ride_length_mean_casual <- mean(as.numeric(casual_df$ride_length))
ride_length_max_casual <- max(as.numeric(casual_df$ride_length))
ride_length_min_casual <- min(as.numeric(casual_df$ride_length))
# Create a data frame
ride_length_metrics <- data.frame(
Metric = c("Mean", "Max", "Min"),
Member = round(c(ride_length_mean_member, ride_length_max_member, ride_length_min_member), 2),
Casual = round(c(ride_length_mean_casual, ride_length_max_casual, ride_length_min_casual), 2)
)
# Print the data frame
print(ride_length_metrics)
# Group by date and count the number of rides each day per group
daily_counts_member <- member_df %>% group_by(started_date) %>% summarise(count = n())
daily_counts_casual <- casual_df %>% group_by(started_date) %>% summarise(count = n())
# super line graph for members
ggplot(daily_counts_member, aes(x = started_date, y = count)) +
geom_line(color = "darkblue") +
geom_smooth(method = "loess", se = FALSE, color = "red", linetype="dashed") +
labs(x = "Date",
y = "Number of Rentals",
title = "Daily Bike Rentals for Members",
subtitle = "With trend line",
caption = "Source: https://divvybikes.com/system-data") +
theme_minimal() +
theme(plot.title = element_text(face = "bold", size = 20),
plot.subtitle = element_text(face = "italic", size = 12),
axis.title = element_text(face = "bold", size = 14),
axis.text = element_text(size = 12))
# super line graph for casual
ggplot(daily_counts_casual, aes(x = started_date, y = count)) +
geom_line(color = "darkblue") +
geom_smooth(method = "loess", se = FALSE, color = "red", linetype="dashed") +
labs(x = "Date",
y = "Number of Rentals",
title = "Daily Bike Rentals for Casual",
subtitle = "With trend line",
caption = "Source: https://divvybikes.com/system-data") +
theme_minimal() +
theme(plot.title = element_text(face = "bold", size = 20),
plot.subtitle = element_text(face = "italic", size = 12),
axis.title = element_text(face = "bold", size = 14),
axis.text = element_text(size = 12))
# Calculate the counts
member_counts <- member_df %>% group_by(rideable_type) %>% summarise(count = n())
casual_counts <- casual_df %>% group_by(rideable_type) %>% summarise(count = n())
# Combine the data into one dataframe
bike_pref_df <- rbind(
mutate(member_counts, user_type = "Member"),
mutate(casual_counts, user_type = "Casual")
)
# Plot
ggplot(bike_pref_df, aes(x = rideable_type, y = count, fill = user_type)) +
geom_bar(stat = "identity", position = "dodge") +
geom_text(aes(label=count), vjust=-0.3, position = position_dodge(0.9)) +
labs(x = "Bike Type", y = "Count", fill = "User Type", title = "Bike Type Preference by User Type") +
theme_minimal()
# Extract hour and day of week from started_at for members
member_df <- member_df %>%
mutate(hour = hour(started_at),
day = wday(started_at, label = TRUE)) %>%
group_by(hour, day) %>%
summarise(freq = n())
# Extract hour and day of week from started_at for casual customers
casual_df <- casual_df %>%
mutate(hour = hour(started_at),
day = wday(started_at, label = TRUE)) %>%
group_by(hour, day) %>%
summarise(freq = n())
# Create the heatmap for member rentals
ggplot(member_df, aes(x = day, y = hour, fill = freq)) +
geom_tile(color = "white") +
scale_fill_gradient(low = "light blue", high = "dark blue") +
labs(title = "Member Rentals by Time and Day of Week",
x = "Day of Week",
y = "Hour of Day",
fill = "Frequency") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust=1))
# Create the heatmap for member rentals
ggplot(casual_df, aes(x = day, y = hour, fill = freq)) +
geom_tile(color = "white") +
scale_fill_gradient(low = "light blue", high = "dark blue") +
labs(title = "Casual Rentals by Time and Day of Week",
x = "Day of Week",
y = "Hour of Day",
fill = "Frequency") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 45, vjust = 1, hjust=1))
# Group by date, user type and bike type to get counts
daily_bike_counts <- combined_data_df %>%
group_by(started_date, member_casual, rideable_type) %>%
summarise(count = n()) %>%
ungroup()
# Now we'll split the data into two dataframes for member and casual
member_daily_bike_counts <- daily_bike_counts %>%
filter(member_casual == "member")
casual_daily_bike_counts <- daily_bike_counts %>%
filter(member_casual == "casual")
# Define our color palette
color_palette <- c("classic_bike" = "blue", "electric_bike" = "red", "docked_bike" = "green")
# Plot for Members
ggplot(member_daily_bike_counts, aes(x = started_date, y = count, color = rideable_type)) +
geom_line() +
scale_color_manual(values = color_palette) +
labs(x = "Date", y = "Number of Rentals",
color = "Bike Type",
title = "Daily Bike Type Preferences of Members") +
theme_minimal()
# Plot for Casual users
ggplot(casual_daily_bike_counts, aes(x = started_date, y = count, color = rideable_type)) +
geom_line() +
scale_color_manual(values = color_palette) +
labs(x = "Date", y = "Number of Rentals",
color = "Bike Type",
title = "Daily Bike Type Preferences of Casual Users") +
theme_minimal()
