Goal — Compare casual vs member rider behavior to identify usage patterns and drive membership growth. Stack — R (tidyverse, dplyr, ggplot2, lubridate, janitor), Tableau Public. Data — 12 months of Divvy/Cyclistic trip data (~4.1M rides).

At a glance

  • Downloaded and consolidated 12 monthly CSVs into one large dataset.
  • Cleaned and transformed records (removed invalid durations, fixed datetimes, dropped NAs).
  • Engineered features: ride_length in minutes, day_of_week, hour, month, and season.
  • Created visualizations showing usage volume, time-of-day trends, and ride duration by type.
  • Published interactive dashboards to Tableau for stakeholder use.

Data & preparation

  • Standardized all monthly datasets using janitor::clean_names() for consistency.
  • Parsed started_at and ended_at with lubridate, calculated ride durations.
  • Applied filters to exclude rides <1 minute, records missing member_casual or station data.
  • Added derived features to support trend and behavior segmentation.

Core import & clean (R):

library(tidyverse)
library(lubridate)
library(janitor)

# Import all monthly CSVs
datasets <- list.files("data_raw", pattern = "\\.csv$", full.names = TRUE)

# Clean and stack into one dataframe
all_trips <- datasets %>%
  map_df(read_csv) %>%
  clean_names() %>%
  mutate(
    started_at = ymd_hms(started_at),
    ended_at = ymd_hms(ended_at),
    ride_length = as.numeric(difftime(ended_at, started_at, units = "mins")),
    day_of_week = wday(started_at, label = TRUE),
    hour = hour(started_at),
    month = month(started_at, label = TRUE)
  ) %>%
  filter(ride_length > 1, !is.na(member_casual)) %>%
  drop_na()

Summary statistics

Using dplyr, I computed aggregate metrics for both rider groups:

summary_stats <- all_trips %>%
  group_by(member_casual) %>%
  summarise(
    avg_ride_length = mean(ride_length),
    median_ride_length = median(ride_length),
    max_ride_length = max(ride_length),
    min_ride_length = min(ride_length),
    ride_count = n()
  )
  • Casual riders take longer rides (~23.6 mins on average).
  • Members ride more frequently, but for shorter durations (~12.6 mins).
  • Casual rides peak on weekends, members ride more during weekdays.

Visualizations & insights

🚲 Ride Volume by Day of Week

  • Casual users spike Saturday–Sunday.
  • Members ride most during weekdays (commuting behavior).

⏱ Average Ride Duration by Day

  • Casual riders take longer weekend rides.
  • Members ride shorter distances but consistently throughout the week.

🕒 Hourly Usage Heatmap

  • Casual: Afternoon/evening use.
  • Members: Clear morning and evening rush hour trends.

All visuals were developed using ggplot2 in R and mirrored in Tableau for stakeholder presentation.

Tableau Dashboard

🔗 View Dashboard

Filters included:

  • Rider type (casual vs member)
  • Day of week
  • Hour of day
  • Ride duration range

KPIs displayed:

  • Total rides: 4.1M+
  • Avg. ride duration by rider type
  • Peak usage windows by day/hour
  • Seasonal and monthly ride trends

Strategic Recommendations

  • Offer weekend ride promos to casual riders.
  • Promote membership benefits after 3–5 casual rides.
  • Use app push notifications to recommend plans based on user patterns.

Deliverables

  • Cleaned and annotated R scripts for data import, transformation, and plotting.
  • Tableau dashboards with rider trends and KPIs.
  • GitHub repository with README and report.

Updated: