This project uses the 2017 bike trips data from Capital Bike Share
- The data contains 4 csv files representing trips data for four quarters.
- Data has several features like
trip start date and time, trip end date and time, start station id and name, end station id and name, bike number and member type. - Analysis is done using Univariate and Multi-variate analysis of the features and creating plots.
- Trend of no. of trips each day over the year and in each quarter
- From which Stations most rides are booked
- Which bikes has taken more no. of rides during the quarters.
- Which start stations and end stations have high no. of members or casual riders.
- Which type of riders have longest rides.
- After the analysis, I tried to predict the member type on the basis of the ride data.
- Random Forests was used with a f1 score of 0.71 on training set and 0.69 on test set.
- Still there is a room of improvement using hyperparameter tuning.