elsebasmar/london-bss

Prediction of Bike Flow

Project description

This project built a prediction model of bike flow in a bike-sharing system to estimate the number of bicycles and docks required in each station at a given point.

Under the hood: all things that happen behind the screen

Step 1: Collecting data

Inputting:

• All Santander trips between each station
• All major events, bank holidays, and disruptions
• Historical weather data

Step 2: Model

Using a SARIMAX that models:

• seasonality
• exogenous variables to take into account additional factors

Step 3: Prediction

Returning a prediction of:

• available bikes at the origin station
• available docks at the destination station
• recommendation of alternative route if necessary

Website(built using streamlit)

On our website:

Enter input information, including an origin, a destination stations and a departure time.

Getting the prediction, including the number of bikes and docks available at the stations(origin and destiantion), trip duration, weather and closest bike stations.