How Can Machine Learning Help to Predict Changes in Size of Atlantic Herring

Explore a conference talk that demonstrates how machine learning techniques can be applied to predict changes in Atlantic herring populations. Learn about the implementation of Python libraries such as Pandas, NumPy, and SciKit-learn to analyze long-term biological data from commercial fisheries. Discover how Gradient Boosting Regression Trees are used to identify key factors influencing the decline in size and weight of Atlantic herring in the Celtic Sea since the mid-1980s. Gain insights into the importance of various environmental and anthropogenic variables, including Atlantic multidecadal oscillation, sea surface temperature, salinity, wind, zooplankton abundance, and fishing pressure. Understand the relevance of this analysis for conservation efforts and sustainable fisheries management, promoting species resistance and resilience. Follow the speaker's journey through problem definition, formal specification, parameter selection, variable analysis, and interpretation of results using partial dependence plots and interaction plots. Conclude with a demonstration addressing multicollinearity issues and discussing final parameters and early stopping techniques in the machine learning model.

Introduction
The problem
Advantages and disadvantages
Formal specification
Parameters
Variables
Partial dependence plots
Interaction plots
Conclusion
Demo
Multicollinearity
Final parameters
Early stopping