1. Ask Phase

1.1 Business Task

The primary objective of this project is to analyze Cyclistic’s historical bike trip data to identify, quantify, and understand the distinct behavioral patterns and usage differences between two core customer segments: Annual Members and Casual Riders. These data-driven insights will directly inform targeted marketing strategies aimed at converting high-value casual riders into dedicated annual members.

1.2 Core Analytical Question

How do annual members and casual riders use Cyclistic bikes differently?

1.3 Key Stakeholders

  • Lily Moreno: Director of Marketing and manager responsible for the development of promotional campaigns across digital channels.

  • Cyclistic Marketing Analytics Team: Data analysts responsible for collecting, analyzing, and reporting strategic data.

  • Cyclistic Executive Team: A detail-oriented leadership group that determines whether to approve the recommended marketing initiatives.

2. Prepare Phase

2.1 Data Sources & Storage

The analysis utilizes Cyclistic’s historical trip data covering the previous 12 months. The underlying public datasets have been generated and made available by Motivate International Inc. under an open data license.

2.2 Data Privacy Guardrails

In strict compliance with data-privacy standards, all personally identifiable information (PII) of riders is completely omitted from the source data. This prevents tracking individual purchase history, credit card connections, or localized residential status.

2.3 Environment Setup & Data Ingestion

The raw batch files are read from the local repository and combined cleanly into a single dataframe via a mapping function.

# Identify all raw CSV trip datasets within the target directory
files <- list.files(
  path = "data_raw",
  pattern = "*.csv",
  full.names = TRUE
)

# Batch import and bind all files into a unified dataframe
bike_data <- files %>%
  map_df(read_csv)

3. Process Phase

The dataset underwent structural audits and automated cleaning to safeguard downstream data integrity.

3.1 Metadata Standardization & Auditing

# Standardize column naming schemas to snake_case format
bike_data <- bike_data %>%
  clean_names()

# Audit missing value counts across variables
colSums(is.na(bike_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 
##            1146841            1146841            1199356            1199356 
##          start_lat          start_lng            end_lat            end_lng 
##                  0                  0               5517               5517 
##      member_casual 
##                  0

3.2 Deduplication & Anomaly Mitigation

Redundant rows are eliminated. Note that manual timestamp parsing using ymd_hms() is bypassed because the read_csv parser automatically ingested the temporal vectors as standard dttm (datetime) data types.

# Drop perfect duplicate entries across the dataframe
bike_data <- bike_data %>%
  distinct()

3.3 Derived Metrics

# Calculate primary trip lengths and pull distinct ordered time components
bike_data <- bike_data %>%
  mutate(
    ride_length = as.numeric(
      difftime(ended_at, started_at, units = "mins")
    ),
    day_of_week = weekdays(started_at),
    day_of_week = factor(
      day_of_week, 
      levels = c("Sunday", "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday")
    ),
    month = month(started_at, label = TRUE),
    hour = hour(started_at)
  )

3.4 Data Validation Filtering

Trips with non-positive or zero-second calculations are removed as operational anomalies (e.g., maintenance testing, docking errors) to prevent analytical bias.

# Filter rows to include valid customer trip data only
bike_data <- bike_data %>%
  filter(ride_length > 0)

4. Analyze Phase

Descriptive statistical evaluations are computed across customer segments to highlight trends.

4.1 Grouped Usage and Baseline Volumetrics

## Calculate absolute trip counts per user group
bike_data %>%
  count(member_casual)
## # A tibble: 2 × 2
##   member_casual       n
##   <chr>           <int>
## 1 casual        1991924
## 2 member        3531283
## Calculate high-level centralized dispersion parameters for trip length
bike_data %>%
  group_by(member_casual) %>%
  summarise(
    avg_ride = mean(ride_length)
  )
## # A tibble: 2 × 2
##   member_casual avg_ride
##   <chr>            <dbl>
## 1 casual            22.9
## 2 member            12.3
## Extract expanded summary metrics for granular distribution analysis
bike_data %>%
  group_by(member_casual) %>%
  summarise(
    mean = mean(ride_length),
    median = median(ride_length),
    max = max(ride_length),
    min = min(ride_length)
  )
## # A tibble: 2 × 5
##   member_casual  mean median   max      min
##   <chr>         <dbl>  <dbl> <dbl>    <dbl>
## 1 casual         22.9  11.5  1575. 0.000767
## 2 member         12.3   8.58 1500. 0.00130

4.2 Grouped Temporal Aggregations

## Generate cumulative usage volume by day of the week
rides_day <- bike_data %>%
  group_by(member_casual, day_of_week) %>%
  summarise(
    rides = n(),
    .groups = "drop"
  )

## Generate average trip duration matrix by day of the week
duration_day <- bike_data %>%
  group_by(member_casual, day_of_week) %>%
  summarise(
    avg_duration = mean(ride_length),
    .groups = "drop"
  )

## Generate monthly trip volumes to identify cyclical season trends
rides_month <- bike_data %>%
  group_by(member_casual, month) %>%
  summarise(
    rides = n(),
    .groups = "drop"
  )

5. Share Phase

The following executive-ready visualizations apply proper contrast, scaling, and labeling to maximize clarity for strategic decision-makers.

5.1 Absolute Ride Count Scale Evaluation

This plot visualizes total operational volume across customer types.

ggplot(bike_data, aes(x = member_casual, fill = member_casual)) +
  geom_bar() +
  scale_y_continuous(labels = label_number(scale = 1e-6, suffix = "M")) +
  labs(
    title = "Number of Rides by User Type",
    x = "User Category",
    y = "Ride Count (Millions)",
    caption = "Cyclistic Trip Data, 2025"
  ) +
  theme_minimal() +
  theme(
    axis.title = element_text(face = "bold"),
    legend.position = "none",
    plot.title = element_text(face = "bold")
  )

5.2 Average Operational Duration Comparisons

This visualization isolates differences in how long casual users ride compared to annual members.

bike_data %>%
  group_by(member_casual) %>%
  summarise(avg_ride = mean(ride_length)) %>%
  ggplot(aes(x = member_casual, y = avg_ride, fill = member_casual)) +
  geom_col() +
  labs(
    title = "Average Ride Duration: Casual vs Member",
    x = "User Category",
    y = "Average Ride Duration (Minutes)",
    caption = "Cyclistic Trip Data, 2025"
  ) +
  theme_minimal() +
  theme(
    legend.position = "none",
    plot.title = element_text(face = "bold"),
    axis.title = element_text(face = "bold")
  )

5.3 Weekly Demand Distributions

This chart uncovers behavioral adjustments across days of the week, helping isolate weekend versus weekday habits.

ggplot(rides_day, aes(x = day_of_week, y = rides, fill = member_casual)) +
  geom_col(position = "dodge") +
  scale_y_continuous(labels = label_number(scale = 1e-6, suffix = "M")) +
  labs(
    title = "Rides by Day of Week",
    x = "Day of Week",
    y = "Rides (Millions)",
    caption = "Cyclistic Trip Data, 2025",
    fill = "User Type"
  ) +
  theme_minimal() +
  theme(
    legend.position = "right",
    plot.title = element_text(face = "bold"),
    axis.title = element_text(face = "bold")
  )

5.4 Macro Seasonal/Monthly Trend Adjustments

This plot reviews long-term seasonal trends throughout the year.

ggplot(rides_month, aes(x = month, y = rides, fill = member_casual)) +
  geom_col(position = "dodge") +
  scale_y_continuous(labels = label_number(scale = 1e-6, suffix = "M")) +
  labs(
    title = "Monthly Ride Trend",
    x = "Month",
    y = "Rides (Millions)",
    fill = "User Type"
  ) +
  theme_minimal() +
  theme(
    legend.position = "right",
    plot.title = element_text(face = "bold"),
    axis.title = element_text(face = "bold")
  )

6. Act Phase

6.1 Key Insights Summary

  1. Volume vs. Duration: Annual members log a higher absolute volume of rides, but casual riders stay on the bikes significantly longer per trip on average.

  2. Temporal Footprints: Annual members show strong consistency during weekdays (aligned with professional commuting hours), while casual user demand spikes significantly on weekends for leisure activities.

  3. Seasonal Factors: Both user segments experience sharp contractions in demand during the winter months, with peak usage concentrated in the summer.

6.2 Top 3 Strategic Recommendations

  1. Introduce Weekend-Only Memberships: Design a specialized annual membership package that targets casual users who primarily ride on Saturdays and Sundays.

  2. Target High-Duration Leisure Hotspots: Launch targeted seasonal marketing campaigns around parks and waterfront routes during the summer, emphasizing the financial savings of converting to an annual membership for long-duration rides.

  3. Commuter Incentives for Seasonal Riders: Use digital media channels to showcase the cost-effectiveness and health benefits of using Cyclistic for daily work commutes, specifically aiming to convert casual weekday riders before the summer peak.