In September, 3 expert scientists from Symrise will take part in Agrostat congress. During their lectures, they will demonstrate how the relevant use of statistical methodologies can help develop palatable, safe, and profitable pet food.
Application of unfolding analysis to palatability versus test
Agrostat | September 4th, 2024 | 4:30 pm | English
Julien Rogues - Data scientist and Artificial Intelligence expert for Symrise Pet Food
The Bradley-Terry model is a well-known method to assess pet food preferences based on paired comparison such as two-bowl tests. However, while this method provides information on the overall appreciation level of a pet food, it may overlook individual preferences.
In his talk, Julien will show how the unfolding analysis method can capture this disparity. Applied alongside with Bradley-Terry model, unfolding analysis can reveal individual variations and preference clusters, offering a detailed understanding of pet preferences.
A Design of Experiments approach to improve the antibacterial effect of protein hydrolysates
Agrostat | September 5th, 2024 | 11:30 am | English
Maxime Fuduche - Microbiology R&D Project Manager for Symrise's Food Protection Platform
Microbial contamination of food and pet food represent a major risk to health, driving the development of effective protection strategies. Bioactive peptides obtained from protein hydrolysis have demonstrated interesting antibacterial properties. However, their efficacy strongly depends on process conditions and pathogen targeted.
During his presentation, Maxime will show how the combination of a fractional factorial design and a Box-Behnken design was successfully used to demonstrate the inhibitory effects of four protein hydrolysates on key foodborne pathogens, understand the impact of the conditions of hydrolysis such as temperature and pH on their efficacy, and determine the ideal conditions to optimize their antibacterial performance according to pathogen.
Time series prediction for raw material prices
Agrostat | September 5th, 2024 | 3:00 pm | English
Pierre Barbe, Data scientist for Symrise Pet food
Most purchasing teams members would like to predict the future. Indeed, envisioning raw materials price movements could definitely enable companies to mitigate risks associated with price volatility.
In his lecture, Pierre will compare the effectiveness of several statistical methods, including machine learning and parametric models, for predicting pet food ingredients price time series based on trade volumes, macroeconomic indicators, ingredient drivers’ prices and other relevant exogenous variables.