1. Goal: Understand variation in bike ridership
2. Useful for company to allocate bikes more efficiently
3. Optimizing the number of bikes that are needed in specific conditions
4. Identifying best times for maintenance work
5. Ensure best possible service to customers

1. Data source: https://archive.ics.uci.edu/ml/datasets/Bike+Sharing+Dataset
2. 17379 samples
3. 14 predictors, containing time, weather and seasonal information
4. Response variable: Total bike count

1. Are there more rentals in warm weather?
2. Do more people ride bikes on holidays or working days?
3. Are there certain peak hours, days or months?

1. On the y axis we have the total bike count
2. Cycle between different predictors on the x axis to understand how bike rentals vary
3. Sample size slider to understand patterns more closely
4. Code snippet getting data and using to plot

    hour = read.csv("hour.csv")
    dataset <- reactive({
        hour[sample(nrow(hour), input$sampleSize),]
    })
    p <- ggplot(dataset(), aes_string(x=input$x, y=input$y)) + 
            geom_point()

1. More rentals in 2012 than 2011
2. Peak in rentals between 8-9am and 5-7pm
3. Maximum rentals in warm weather