Applied Data Science with R - Capstone Project Summary

• Objective: Predict the number of bikes rented based on weather conditions in urban areas.

• Regression Model Insights:

  • Positive correlation found between demand and temperature.
  • Negative correlation observed between humidity and rainfall.

• Methodology:

  • Utilized diverse regression models for prediction.
  • An interactive map enhances visualization of estimated bike demand.

• Data Collection:

  • Comprehensive data collection from global bike-sharing systems, OpenWeather APIs, and cloud storage.
  • Specifics gathered from Seoul's bike-sharing systems, including major cities' information, weather details, and hourly bike rentals.

• Data Wrangling:

  • Stringr and regular expressions for column standardization and numeric value extraction.
  • Dplyr used for handling missing values, creating dummy variables, and normalizing data.

• Exploratory Data Analysis (EDA):

  • SQL queries and data visualization techniques employed.
  • Insights obtained on bike-sharing systems, weather, and city-specific conditions.

• Predictive Analysis:

  • Multiple regression models evaluated using RMSE and RSQ values.
  • Models include baseline linear regression, higher-order polynomials, and penalties like Lasso, Ridge, and Elastic Net.

• R Shiny Dashboard:

  • Interactive interface for exploring bike demand predictions, temperature forecasts, and humidity-rental demand relationships across cities.

• Conclusion:

  • Valuable insights into bike-sharing demand and applied data science with R showcased.
  • Future avenues include exploring additional weather conditions and analyzing policy impacts on bike-sharing demand.