- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Complementarity of experimental plans and Machine Learning in the service of business expertise
The aim of the study is to study a simplex mixture of three raw materials and characterize it in terms of price, viscosities and stabilities at different conditions.
Due to a difficulty in postulating an a priori model and a possible area of instability of the formulas which could compromise the plan and the subsequent analysis of the results (no measurable response in the event of instability), a plan Space-Filling type with excluded zone is carried out.
A first modeling with different Machine Learning type models (SVM, Gaussian Process) is carried out, but certain areas of the experimental space are poorly represented due to missing values for viscosity (too low viscosity or instability of some formulations).
Using information from business expertise, and the choice of imputation of local values by K-Nearest Neighbors, the modeling can be corrected and provide satisfactory results, providing a better representation and understanding of the formula space .
- Chapters
- descriptions off, selected
- captions settings, opens captions settings dialog
- captions off, selected
This is a modal window.
Beginning of dialog window. Escape will cancel and close the window.
End of dialog window.
This is a modal window. This modal can be closed by pressing the Escape key or activating the close button.
Biography
Passionate about data science and chemistry, my goal is to combine business expertise, experimental designs and data analysis to help laboratories leverage their historical data, effectively plan their next experiments, make better decisions and to unravel some of the mysteries of chemistry.
Rational and data-driven, I am ready to share my passion and help make data literacy accessible to everyone. Curious, I love reading and learning about innovations, technology, science, chemistry, statistics, data science and analytics.
Area of interest: Innovation, Digitalization, Design of Experiments, Statistics, Machine Learning, Data Analysis, Data Visualization and Chemoinformatics.
Company presentation
For more than a century, L'Oréal has devoted itself solely to one business: beauty.
L'Oréal wants to bring beauty to all people. The goal is to win over another one billion consumers around the world by innovating and creating cosmetic and skincare products that meet the infinite diversity of their beauty needs and desires.
This post originally written in French and has been translated for your convenience. When you reply, it will also be translated back to French .
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: Complementarity of experimental plans and Machine Learning in the service of business expertise
List of resources and additional information available in the presentation:
- General Machine Learning and algorithms :
- Guide to AI algorithms - MFML Part 4 playlist by Cassie Kozyrkov
- Bias Variance Tradeoff (mlu-explain.github.io) the explanation of the bias-variance trade-off in ML to avoid overfitting/underfitting
- Additional information on Support Vector Machines :
- Overview of Support Vector Machines (jmp.com)
- Support Vector Machines (SVM): An Intuitive Explanation | by Tasmay Pankaj Tibrewal
- SVM with polynomial kernel visualization
- Support Vector Machines Part 1 (of 3): Main Ideas!!! -StatQuest
- Support Vector Machines Part 2: The Polynomial Kernel (Part 2 of 3) - StatQuest
- Support Vector Machines Part 3: The Radial (RBF) Kernel (Part 3 of 3) - StatQuest
I remain available for any questions, remarks or comments!
This post originally written in French and has been translated for your convenience. When you reply, it will also be translated back to French .