Identification of the optimal location of co-working spaces using clustering methods. Notebook and source code repository of the solution use case.
The solution was developed to support decision makers in optimize work travels and reduce costs and incurred by employee of Regione Emilia-Romagna and Agenzia per il Lavoro (Regione Emilia-Romagna).
How the repository is organised:
- notebooks: notebooks used to develop solution
- docs: any report or technical documentation of solution implementation
The implementation uses a distributed computation engine (Apache Spark) to speed-up performance on simulation and external API (Open Source Routing Machine) to obtain distance and travel time between workplaces and employees' addresses. Simple data transformation are performed in order to re-shape data in a useful way using python most famous library pandas. Clustering analysis is performed with SciKit-Learn python library.
Output can be consumed via every common BI tool.
A list of requirements is attached to this repository in order to run notebooks inside a Python environment. We suggest install libraries within a virtual enviroment.
$ python -m venv myvenv
$ source myvenv/bin/activate
$ pip install -r requirements.txt
In order to adapt the solution to your business case notebook have to be excuted.
- 01_Ingest.ipynb notebook shows how to ingest, clean, format and standardise data in order to perform the analysis.
- 02_Cluster.ipynb notebook perform cluster analysis of employees' geo-location and choose optimal configuration.
- 03_Simulate.ipynb notebook calcutate distances and duration for employee/workplace pairs.
Contact the authors for informations about the project.
Andrea Iacobucci, Serena Borsari, Luca Bonafede
The project is still active and under development for future improvements.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent