Technology Watch

Description

This project was carried out as part of my computer engineering studies for my company Thuasne. The purpose of this watch is to discover and apply the principles of technological watch and share the knowledge acquired.

Technological watch, or digital watch, consists of systematically gathering information about the latest techniques and especially about their commercial availability. Technological watch for a company is useful for:

• Preparing a business project
• Researching new tools
• Tracking new trends
• Marketing
• Improving productivity

It is a research and development tool that allows for the collection, analysis, and dissemination of information deemed relevant, following an in-depth study of a new technology.

This project resulted in a report and a presentation.

Objectives

This technological watch should provide a clearer view of the data analysis challenges. For about twenty years, it has become clear to companies how important data, and more particularly its analysis, is to the point of constituting a real competitive advantage.

It is therefore a current practice today, especially in the health field. However, the implementation of connected medical devices at Thuasne is particularly recent. Thus, the infrastructure and processes related to health data had to be considered and developed (data structure, anonymization…). It was therefore a bit early to think about implementing data analysis under these conditions. However, thanks to the development of the support platform and all the IoT infrastructure deployed on Azure, we can now exploit the data under good conditions.

The purpose of this technological watch is therefore to have an overview of what can be done with the health data we have, but also of the entire process required to achieve it. Here, we are talking about the technologies used, the necessary infrastructures and workforce, but also the cost of the entire process in order to make a decision on the possible application of this new process within Thuasne.

Thus, the issues I have identified are:

1. What is data analysis?
2. How can data analysis be beneficial for Thuasne?
3. What are the most suitable means and tools for implementing data analysis for Thuasne?

Methodology

To successfully carry out this project, I had to implement a rigorous work methodology. Indeed, since the topic was broad, I first had to select the key information and restrict the range of information I was going to keep and address in this watch. In addition, I was able to read a lot of research papers and content on the subject. Therefore, it was necessary to keep everything in one place, annotate the important passages, and especially verify my sources.

Technologies Used

I used various tools during this project:

• Notion
• Google Drive
• Zotero
• Web Highlight
• Kaggle
• Google Scholar

Experimentation

Finally, I was able to select two interesting technologies for use in the context of my company. These are Python and R. To compare them, I conducted a study of a dataset with both languages. All of this within two notebooks available right here: Python & R.

Conclusion

This technological watch project was very instructive and enriching for me. However, its realization posed many problems for me. Indeed, the subject is the analysis of health data. With all the resources available on the Internet, the impressive number of recent innovations in the field, and the number of concepts behind each solution, it was really difficult to restrict myself to the dozen pages required by the school.

I tend to always write a little too much, but this is because I appreciate that all my content is clear and without ambiguity. This project was therefore a very good exercise in extreme synthesis for me. There are many concepts that I could not develop and define, such as machine learning for example.

Furthermore, I had to restrict myself in terms of implementation in order to have something easy to understand and sufficiently complete. For this, I had to skip some steps that allow for a good quality data analysis, such as normalization for example.

I was able to learn a lot from the public resources of different people on Kaggle who presented their approach to data analysis. Thus, I would have liked to do a more complex and complete experimentation.