This repository refers to the implementation and the structure of a big data system built to predict the usage and the status, for a given day and a specific time slot, of e-charging stations and plugs in the Italian region of Trentino Alto Adige. The service, through a simple interface, is capable of providing to a user, who could be either a private driver or a maintenance technician, with the possibility to predict the status of a station up to the next 14 days.
-
retrievalsfolder: it contains three files which refer respectively to three different data collection from three distinct API calls and include code for a preparatory data processing phase. -
utils/random_forest.py: this file contains the code through which we instantiated our model, we trained and tested it, eventually saving the output in a .joblib file. -
utils/redis_keys.py: this code allows us to retrieve data from Big Query and populate our local Redis instance with key-value pairs regarding static data of our stations. -
docker-appfolder: in this folder you can find all the required files needed to build the Docker image, referring to the web application. Please note that the .joblib file contained in themodelsubfolder refers to a reduced version of the full model, this has been done because the full model's size is too large to be pushed on GitHub. -
docker-fullfolder: this folder includes the docker-compose.yml file which is necessary to run our full solution, composed by both the web application and the Redis instance.
- Pull the image on GitHub Packages
- Download docker-full: docker-full
- Unzip it and run
docker-compose upinside the folder (it takes around 1 minute for the service to be up and running) - Enter http://localhost:5000/ in a browser to see the application running