Chicago-Bike-Share-Company-Project

Introduction

This project is a part of Google Data Analytics case study, and the data keeps updated periodically by the provider. Cyclistic is a bike-share company in Chicago. In 2016, the company launched a successful bike-share offering with 5,824 bicycles and 692 docking stations. They primarily provides 3 types of bikes: classic, electric and docked bikes. The director of the marketing team wants to maximize company's profit by converting more casual riders to members.

Goal

My goal is to design a new marketing campaign to convert riders into annual subscription by providing actionable recommendations, backed up with compelling data and data visualizations. Casual riders are defined as customers who purchased a single-ride or a full-day pass. Annual members are defined as customers who purchased annual memberships.

In order to design a marketing strategy to convert more casual users, it's important to ask:

1. How does user behavior differentiate between casual riders and members?
2. Why would casual riders buy Cyclistic annual memberships?
3. How can Cyclistic use digital media to influence casual riders to become annual members?

Data Source

Motivate International Inc. provides with the data and is available under this license. The data used for analysis is sourced from Cyclistic's bike trip records and I selected data from Jan 2023 until Jun 2023 (6 Months) for my analysis to better understand trend and user difference. Each dataset includes the following 13 fields: ride_id, rideable_type, started_at, ended_at, start_station_name, start_station_id, end_station_name, end_station_id, start_lat, start_lng, end_lat, end_lng, member_casual

Data and Data Privacy

1. Riliability and Originality: The data provided is internal data from the company and is owned by the City available to the public, and also there's a huge sample data with more than 5 million trips per year. So this should be original and reliable source of data.
2. Comprehensiveness: In each dataset, it includes information about ride_id for each trip, ride timestamp(start & end), station name (start & end), geographic data including latitude and longitude of each ride, and user status(member or casual)
3. Current: The data is continuous to updata monthly
4. Cited: Yes

Data Limitation

1. We don't have access to personally indentifiable data, so we couldn't identify if the same rider who rides more than once as a casual rider or a member or which age group is more preferrable to ride a bike.
2. The datasets have many NULL values in the start_station_id and end_station_id, especially under electric_bike category, hence it's hard to draw a conclustion of the most visited station by electric_bike riders.

Procedures

1. Data Wrangling (See my code here)

   • Filter out invalid entries that started_at later than or equal to ended_at, and delete entries from the dataset
   • Create two columns: duration represents the length of each ride; day_of_week represents which day of week the ride happens
   • Filter out bike duration greater than 12 hours, it might cause by forgetting to lock the bikes accidently
   • Format data by trimming lag/lng data whitespaces at the beginning and at the end, and round lag/lng to 6 decimals since it's good enough to see a person
   • Finally, I created a view, formatted_bike_data, with rides under 12 hours, and with correct data format

2. Data Analysis (See my code here)

   For this project, we aim to find out what makes casual users want to become an annual member. We're going to study user behavior between casual users and annual members.

   Usage Patterns Analysis

   • What's the different between ride duration between casual riders and members?
   • What's ride duration differ in the member types?

   Seasonal Trend

   • Is there any seanality trend for the first half of year between casual riders and annual members?

   Day and Time Analysis

   • Identify peak hours and days for annual and casual riders by day of week? by hour?

   Bike Type Preferences

   • What's preference for different bike types between casual and annual members?

Data Visualization

After conducting a thorough analysis of the bike-sharing data, a notable trend emerges: annual members constitute the majority of rides, accounting for a substantial 65%. In contrast, casual users represent a smaller share, contributing only 35% of the total rides.

The figure below shows the total number of rides carried out by Cyclistic members and casual riders of the first half year in 2023.

To learn more about differences between casual riders and annual riders, I want to know how long usually a ride is for the two types of users. This below figure shows casual riders’ average trip length is more than annual member rides

Next, I want to understand seasonal trend in user behavior. I would like to identify if there are certain times of the year, of the day of week, and of the hour of day when casual or annual users are more likely to use the service. This below figure shows for the first half of the year, the peak month of rides for both casual and member is June, and we noticed a upward trend in rides. This finding explains people tend to ride more during faborable weather conditions.

Deep dive in the day of week trend, I found that annual members have different user behavior from casual users. In details, annual members tends to ride more on Tuesday, Wednesday, and Thursday. And this could due to majority companies have a hybrid work schedule, usually 3 days work from the office and 2 days work from home. In contrast, casual riders tend to ride more on weekends than weekdays, and Saturday is the peak day of the week.

To substantiate the hypothesis suggesting that annual members exhibit higher ride frequency on weekdays due to commuting, I conducted a detailed ride analysis by hour. This allowed me to discern patterns and trends in the data, specifically focusing on the hours of the day when members are most active.

From the above figure, we saw annual members show peak bike usage during the morning hours from 8 am to 9 am and in the afternoon from 5 pm to 7 pm. On the other hand, casual members exhibit an upward bike usage from 11am till 6pm, and a downword trend since then. We concluded that during specific hours, annual members have a clear purpose to ride a bike.

In addition, in order to help inform decisions on marketing campaign strategy and bike fleet management, I did analysis on bike preferences.

The analysis reveals the electric_bike is preferred choice for both groups, followed closely by by the classic_bike. However, a notable distinction emerges with docked_bikes, as they are exclusively used by casual riders, with no annual members utilizing them.

Recommendations

In order to design a marketing strategy to convert more casual users into members, we've found that: 1). Casual riders mostly ride during the weekends 2). Average ride length of casual driver is 1.69 times of that of annual member on each ride 3). For both casual and annual riders, as the weather is getting better, there's an upward trend seen in the first half of the year 4). For both casual and annual riders, they prefer classic_bike to other bikes, and also, docked_bikes are only exclusively occupied by casual riders

Based on above observations, we come up with campaigns that: 1). Provide various membership program: instead of providing only 3 types of memberships: one-ride pass, full-day pass and annual members, we should provide more choices. For example: seasonal membership targeting on riders only ride during summer time and fall time, and weekend pass that only allows people to ride on weekends 2). Socia Media Awakeness: We should encourage people to share their riding experience with friends and families on social media by giving them free rides as an incentive 3). Pricing Strategy: We could charge casual riders base on rate/mile since causal riders have more ride lenghth than annaul members , and make them think it's better off just to get an annual membership.