amqsabnur / bike-share-revenue-forecast

Forecast daily bike-share revenue in Jersey City for next 30 days. Tasks include data cleaning, feature engineering, exploratory data analysis, merging datasets, correlation analysis, and implementing the FB Prophet model.

Data

• New York City and Jersey City Bike Share data from Kaggle (https://www.kaggle.com/datasets/akkithetechie/new-york-city-bike-share-dataset/)
• New York City weather data from (https://www.wunderground.com/history/daily/us/ny/new-york-city)

Task

• Forecast the daily revenue for next 30 days
• Complete the forecast for only 7 selected stations of choosing
• Assume the cost of bike rental is $1.00 + $0.10 per minute until the ride lasts

Task breakdown

• Step 1: Data Cleaning, Feature Engineering
  • Clean the bike trip dataset, handling missing data, and dropping unnecessary columns
  • Create additional features like day of the week, hour of the day, public holidays, user age, revenue calculation
• Step 2: Exploratory Data Analysis (EDA)
  • Perform EDA to understand variable distributions, identify outliers, and visualize trends in trip durations
  • Explore the distribution of trip durations and assess their impact on revenue calculations
• Step 3: Data Merging
  • Develop a robust merging strategy, ensuring proper alignment of time periods and handling missing data appropriately
• Step 4: Correlation Analysis
  • Conduct a thorough correlation analysis within both weather and bike trip datasets, identifying features strongly correlated with revenue
• Step 6: Visualizing Full Data
  • Visualize the data
• Step 5: Station-Specific Visualization
  • Select 7 stations
  • Explore dynamic and scalable solutions for station-specific analysis, possibly through functions or loops
• Step 8: Feature Selections
  • Finding the most important features responsible for revenue
• Step 9: FB Prophet Model
  • Implement the FB Prophet model for forecasting daily revenue for January, March, July, and October 2016
  • Focus on the 7 selected stations
  • Assume a cost of bike rental as $1.00 + $0.10 per minute