Creation Review

Type
Creation
State
Successful
End Date of the Review Period

Reviews run for a minimum of one week. The outcome of the review is decided on this date. This is the last day to make comments or ask questions about this review.

Project
Proposal

Eclipse Grela

Tuesday, February 2, 2021 - 06:33 by Juan de Oliveira
This proposal is in the Project Proposal Phase (as defined in the Eclipse Development Process) and is written to declare its intent and scope. We solicit additional participation and input from the community. Please login and add your feedback in the comments section.
Project
Parent Project
Proposal State
Created
Background

s0nar is a solution developed by Improving Metrics to perform IoT data analysis through Artificial Intelligence methods including Machine Learning and LTSM neural networks. 

s0nar has its origin in the experience in Machine Learning based anomaly analysis in Digital Analytics business cases and especially thanks to the participation as a consortium member in the Brain-IoT project, within the H2020 programme of the European Commission. In this project, Improving Metrics was appointed partner responsible for developing the key features of Artificial Intelligence and Machine Learning.

s0nar's vision goes beyond ad-hoc development for Brain-IoT, to become a scalable product specialised in AI-based IoT data analytics. Its open design and key features allow agile and flexible implementations.

s0nar will become Eclipse Grela.

Scope

Eclipse Grela is a highly innovative technological product that automates the entire workflow necessary for the implementation of predictive systems and anomaly detection systems, both on IoT sensor time series. 

Description

Eclipse Grela carries out the automatic versioning of datasets, the Data Science work of data analysis, cleaning and selection and the machine learning work of model adjustment and training. All these phases are configurable, being able to adjust the requests to the specific use case to be solved.

The integration with clients or external systems is simplified, giving access to web endpoints from which datasets can be registered, models can be trained and results reports can be obtained, allowing the aforementioned configurations. The architecture allows the scalability of several clients at the same time.

The functionalities of Eclipse Grela can be divided into two main groups, those aimed at prediction and those aimed at anomaly detection. Both groups are implemented on the same architecture and can be used independently. 

Why Here?

The goal of Eclipse Grela is to offer a platform to make easier to train, deploy and integrate Machine Learning analysis features in IoT time series use cases. The Grela Eclipse community can help us to build a better product together by creating new reusable ML functionalities.

Future Work

Extend use cases, increasing security, stability and making technical improvements that guarantee scalability. The use cases are located in the energy, water, transport infrastructure and industry verticals.

Project Scheduling
  • Promote an initial user community, based on open source licensing of the core of the initial development. (February 2021)
  • Establish multiple proof-of-concepts (PoCs). Intensify initial commercial contacts and be able to test the validity of the product in multiple companies. Initial contacts need to be converted into proofs of concept aimed at solving specific use cases. (March 2021-July 2021)
  • Incorporate learning from the PoCs into the tool, in order to strengthen the behaviour and incorporate functional improvements. Once multiple PoCs have been conducted, first-hand knowledge is obtained to improve the tool and make it more successful. (August 2021-November 2021)
Project Leads
Committers
Manuel Miranda Romero
Initial Contribution

Eclipse Grela (previously s0nar) is an IoT Time-Seris Analytics service  based on a modern technological stack to be deployed in cloud environments, with technologies such as Grafana, Prometheus, RabittMQ, Kubernetes, Celery, Python, Tensorflow, Nvidia or MongoDB. 

Third-party libraries and associated licenses

flask - BSD

flasgger - MIT

flask-mongoengine - MIT

pymongo - Apache

celery - BSD

boto3 - Apache

environs - MIT

pytest - MIT

pytest-flask - MIT

pytest-mock - MIT

pytest-cov - MIT

pandas - BSD

numpy - BSD

matplotlib - BSD

scikit-learn - BSD

pmdarima - MIT

Tensorflow - Apache

Source Repository Type