xxinru / kaggle_bike-sharing-demand

Kaggle Competition: Bike Sharing Demand

• Ranking Top 16.5%
• competition url: https://pse.is/3rryf8

Competition Description:

Bike sharing systems are a means of renting bicycles where the process of obtaining membership, rental, and bike return is automated via a network of kiosk locations throughout a city. Bike sharing systems therefore function as a sensor network, which can be used for studying mobility in a city.

The competition goal:

To predict the total count of bikes rented during each hour covered by the test set, using only information available prior to the rental period.

Prediction step:

1. Exploratory Data Analysis (EDA)
2. Feature Engineering
   • Outliers Detection
   • Feature Transformation
   • Missing Value Imputation
3. Build machine learning models
   • Linear regression
   • Polynomial regression
   • RandomForest regressor
4. Choose model and predict