Introduction
In 2015, the bicycle rental service Seoul Bikes was founded, which aims to promote a sustainable means of transportation for the citizens of South Korea. This system has stations spread across Seoul where it is possible to rent bicycles, which are considered delivery and pick-up points. Using the Seoul Bikes app, the availability of bicycles can be checked, something that facilitates the associated logistical process.
Problem at hands
Using a set of public data relating to bicycle rentals from the company Seoul Bikes, together with the weather conditions that affect this same rental, the objective of this project is to predict the number of bicycles needed to be made available each hour in order to make the service as efficient as possible.
Criteria for Success
Faced with this situation, it is pertinent to take into account the associated risks, namely, the scarcity of bicycles available given the demand at a given collection point. As preventive measures against such risks, it was considered that a minimum margin of available bicycles should be established with the intention of maximizing profit while simultaneously ensuring the availability of the service. In short, the work is considered successful if a model with generalization capacity and minimum prediction error is obtained.
Used tools
To carry out this project we had 5 2nd year Data Science students, a supervising teacher, a dataset from the UCI Machine Learning Repository, personal computers, Jupyter Notebook, Python programming language, Prezi and Django.
Full Report
The full report containing detailed explanations of the developed work, as well as the obtained results is present in the 'Script_Relatorio/Relatório_Grupo12_PACD1.pdf' file.