The purpose of this course is to provide the participants with knowledge

about emerging statistical methods; their functioning, application, and

properties.

The course will feature innovative tools from the field of spatial analysis

and how they could provide relevant insights into the structure of

ecosystems under anthropic pressures, by dissecting across multiple

spatial scales.

This course will cover a range of computational methods applicable for

the analysis of toxicants and environmental sciences’ data.

  • Methods to pattern and cluster biota community according to the

toxic substances.

  • Modelling approaches (e.g. artificial neural networks, classification

models) to predict aquatic ecosystem quality by using the

responses of biota community to changes of the environment. Two

types of algorithm will be presented with relevant application

examples:

  • unsupervised learning models (ordination and clustering)

used for patterning communities according to the effects of

toxic substances,

  • supervised learning models (predictive models) used to

predict community disturbances by anthropogenic impacts,

especially toxic substances.

  • The use of sensitivity analysis techniques to illustrate the

relationships between toxicants and biota.

  • Modelling of structures (e.g. spatial, temporal) using Moran's

eigenvector maps and asymmetric eigenvector maps.

  • Analysis of multiple scale correlative features using the Multiscale

Codependence Analysis.

The methods will be applied on large datasets. The participants will be

given practical examples using the R statistical software.