Analyzing Bike Sharing Demand

This project is analyzing bike sharing demand in the context of ride-hailing apps such as Uber and Ola. Bike-sharing systems have become increasingly popular in urban areas as a sustainable mode of transportation, and this trend aligns with the goals of ride-hailing companies to provide convenient and eco-friendly travel options to their users.

Issue of Interest:

Analyzing bike-sharing demand is important for several reasons. Firstly, it can aid urban planners and policymakers in designing efficient transportation networks. Secondly, bike-sharing companies can use this analysis to optimize fleet management, station placement, and pricing strategies. Lastly, studying bike-sharing demand patterns can provide insights into commuting habits, environmental impact, and public health.

The boxplot with meaningful information is the Hour of the day because it shows the busiest hours are 7AM-8AM and 5 PM-6 PM, which means the users mainly rent bikes to ride to work/school and to return back home at the end of the day. Based on boxplots 4 and 5 (Working Day and Holiday), we see most outliers are in working days. The result makes sense when looking at holidays since all outliers are for non-holiday days. The last boxplot demonstrates the obvious common sense that most users rent bikes when the weather is Clear and Cloudy (1 and 2), and almost no users when is heavy raining or snowing (3).

Techniques/Models:

1.I have used Correlation matrix for feature selection.

2.Removed the outliers from the selected features and target variable.

3.The variables/measures that were used in the analysis datetime,season,holiday,workingday,weather,temp,atemp,humidity,windspeed,casual,registered,count.

4.Trained the model with different ML algorithms like Linear Regression model, Random Forest Regressor model, Gradient Boosting Regressor model and Extreme Gradient Boost Regressor method.

5 Extreme Gradient Boost model(XG Boost) has the least RMSE which is 0.08 (Root Mean Squared Logarithmic Error) we have chosen that one for our prediction and the accuracy for that is 97.1 %

Conclusion:

1.As we embrace the transformative power of Machine Learning for demand prediction, businesses can unlock new possibilities for operational efficiency and customer satisfaction.

2.As cities strive to create cleaner, healthier, and more accessible environments, bike-share systems offer a remarkable solution.

Previous
Previous

Client Scorecard

Next
Next

HR Analytics