Cette publication est également disponible en : Français
How did your collaboration and the idea for your project come about?
We met during our work at the IDYST research group: Geosciences and Knowledge Discovery in Data (GeoKDD) (Federico as a post-doc and Fabian as a PhD student). Our research covered multiple aspects of data analysis: spatial statistics and modeling techniques, machine learning, artificial intelligence, environmental risks and energy potentials modeling.
We quickly realized that these approaches and our expertise had potential for interest outside the academic world. Indeed, private or public sector companies must adapt to climate and environmental issues. To do this, they often have to process complex data. However, these organizations do not always have sufficient financial or human resources to process these data.
Our profile allows us to interact as competent interlocutors on both sustainability and environmental issues environment, as well as on data analysis requiring complex IT models and tools. By leveraging this dual background, we can take concrete actions in favor of the climate, helping companies to be more resilient to climate change or to optimize their activities in more sustainable models.
What does Grinsight propose, what are the projects developed?
Grinsight (a mix between green and insight) aims to offer support to companies, public services or NGOs that wish to assess the environmental impact of their activities or to reorient their operations or business model towards more sustainable solutions.
The support focuses on three areas:
- helping the organization to define the problem it wants to solve and identify the data useful to find solutions; possibly helping to collect these data,
- define a strategic development plan based on the analysis of the data collected; identify the impacts of climate change on the organization and/or the possibilities to reduce its carbon impact,
- develop a risk-management plan to prevent potential risks related to the organization’s future activities.
The field of application is very broad and fits into various sectors of activity. Our service can range from the creation of socio-environmental models for public administrations and NGOs, to the use of satellite and meteorological data to evaluate the impact of global warming on the frequency of hail events in our region.
However, the current objective is not to canvass all directions. Grinsight remains for the moment a secondary activity. Our main activity is in the areas of innovation, research and development in data science. One of Grinsight’s goals is to facilitate the transfer of this academic knowledge to concrete applications within society.
What were the steps to set up the project and to participate in the UNIL Entrepreneurship and Innovation HUB?
When the idea of setting up a start-up matured towards the end of our FGSE course, we looked for levers that would allow us to make it happen. Among several solutions (PACTT, HUB UNIL…) we opted for the UCreate 3 program (HUB UNIL) because it was the one that was the closest to a first step. We wrote our project idea on a few pages and submitted our application in September 2020. Our project was selected among 12 projects (out of more than 60 proposed) for the first round. We then participated in an interview that allowed us to be among the 7 projects finally supported.
UCreate3 provided us with various types of support:
- a sum of CHF 10’000.- which allowed us to develop the visual identity of Grinsight (logo, website etc.)
- courses and personalized coaching follow-up that forced us to free up time to think about our project and to precisely define the business model (which we do not necessarily do spontaneously when we are busy with our daily work). The coach’s contribution throughout the project was very beneficial.
- support of conducting a desirabilitystudy. This consist of contacting potential clients to present the company and its services to them, to see if our proposal arouses any interest. This step was very useful: it taught us to dare to pick up the phone and to formalize our services. We were able to establish a first contact that finally materialized into two mandates the following year.
During the development of the project, we had to explain our objectives to people who were not in the field. This required us to learn how to communicate to a business-oriented audience, which requires a different language than the academic world. We therefore had to popularize our discourse and focus on our services.
What advice would you give to UNIL PhD students /post-doc fellows who are thinking of starting their own business?
The UNIL Entrepreneurship and Innovation HUB, as well as PACTT, are accessible and offer concrete help. Innosuisse also offers a range of tools for all those who want to explore the possibility of transforming their research into an application that could lead to the creation of a company. Whatever the option chosen, if an idea seems interesting, one must go through with it. Even if in the end the company does not succeed or lasts only for a while, you learn a lot during the different steps: better communicating, formalizing ideas for a non-scientific audience, going outside your framework to contact companies etc… Moreover, this experience is enriching. It allowed us to discover facets of ourselves that we will be able to value on a CV.
What is Machine Learning?
“Machine learning” is a family of algorithms, belonging to the field of artificial intelligence that can learn automatically from experience, without the learning rule being explicitly programmed to do so. The basic algorithm does not change, but improves its results as the amount of data presented to the computer becomes larger and larger.
Thesis defended by F. Guignard which gave rise to a monograph published by Springer
- On Spatio-Temporal Data Modelling and Uncertainty Quantification using Machine Learning and Information Theory
Scientific articles at the intersection of data analytics, artificial intelligence, and the environment:
- A novel framework for spatio-temporal prediction of environmental data using deep learning
- Spatio-temporal evolution of global surface temperature distributions