School of Environmental Sciences

Ecological Data Analysis with Machine Learning Methods

This course is part of the programme:
Master’s study programme Environment (2nd level)

Objectives and competences

Introduce the students to the field of ecological data analysis with machine learning methods. The students will gain basic knowledge about data analysis with machine learning methods. Also, they will gain an overview of some widely used machine learning methods. They will be familiarized with case studies that used machine learning to analyse ecological data. Through exercises, the students will get acquainted with some machine learning software packages.

Prerequisites

Completed Bologna B.Sc. or professional type of undergraduate education

Content (Syllabus outline)

1. Introduction to knowledge discovery and machine learning methods (learning decision and regression trees; learning rules; Bayesian classification; nearest-neighbor method; equation discovery)

2. Classes of ecological problems for which machine learning can be used (population dynamics and habitat modeling)

3. Case studies of using machine learning to analyse ecological data (aquatic ecosystems, agriculture, forestry / for example, modelling algal growth in the Lagoon of Venice and Lake Bled, modeling brown bear habitat)

4. Exercises in applying selected machine learning methods to ecological data (demostration/exercises with machine learning software packages)

Intended learning outcomes

Knowledge and understanding:

A student who completes this course successfully will know and understand

General Competences:

  • The process of Knowledge Discovery
  • Machine learning methods, such as decision and regression trees, rules, nearest neighbors methods, equation discovery
  • Selection of an adequate machine learning method to analyze given data
  • Acquaintance with various machine learning software packages

Course Specific Competences:

  • Basic concepts of knowledge discovery and machine learning
  • Overview of most widely used machine learning methods
  • Machine learning methodology: different case studies-different approaches
  • The ability to identify a suitable machine learning method that can be used to analyse given ecological data
  • The ability to apply a machine learning method to a given data
  • The ability to use some machine learning software packages

Readings

A. Fielding, editor. Machine Learning Methods for Ecological Applications. Kluwer, 1999.

S. Dzeroski. Data mining in a nutshell. In S. Dzeroski, N. Lavrac, editors, Relational Data Mining, pages 3-27. Springer, 2001.

S. Dzeroski. [KDD Applications in] Environmental sciences. In W. Klösgen, and J. M. Zytkow, editors. Handbook of Data Mining and Knowledge Discovery, pages 817-830. Oxford University Press, 2002.

A. F. Zuur, E. N. Ieno, and G. M. Smith. Analysing Ecological Data. Springer, 2007.

I. H. Witten, E. Frank, and M. A. Hall. Data Mining: Practical Machine Learning Tools and Techniques. Morgan Kaufmann, 2011.

Assessment

  • Seminar (student project), including a written report and oral defense (100 %)

Lecturer's references

Elected Full Professor in 2010 for the area Computer and Information Science at the Faculty of computer and information science, University of Ljubljana.

  • ŠKERJANEC, Mateja, ATANASOVA, Nataša, ČEREPNALKOSKI, Darko, DŽEROSKI, Sašo. KOMPARE, Boris. (2014) Development of a knowledge library for automated watershed modeling. Environmental Modelling & Software, 54: 60-72. (COBISS-ID= 6485601)
  • KOCEV, Dragi, DŽEROSKI, Sašo. (2013) Habitat modeling with single- and multi-target trees and ensembles. Ecological Informatics, 18: 79-92. (COBISS-ID= 26909735)
  • ČEREPNALKOSKI, Darko, TASHKOVA, Katerina, TODOROVSKI, Ljupčo, ATANASOVA, Nataša, DŽEROSKI, Sašo. (2012). The influence of parameter fitting methods on model structure selection in automated modeling of aquatic ecosystems. Ecological Modeling, 245: 136-166. (COBISS-ID= 26089767)
  • KELLER, Reuben, KOCEV, Dragi, DŽEROSKI, Sašo. (2011). Trait-based risk assessment for invasive species : high performance across diverse taxonomic groups, geographic ranges and machine learning/statistical tools. Diversity and Distribution, 17(3): 451-461. (COBISS-ID= 24674087)
  • EROSKI, Sašo, TODOROVSKI, Ljupčo, editors. Computational Discovery of Scientific Knowledge. Springer, Berlin, 2007. (COBISS-ID= 20974631)

University course code: 2OK029

Year of study: 1

Course principal:

Lecturer:

ECTS: 6

Workload:

  • Lectures: 30 hours
  • Exercises: 15 hours
  • Individual work: 135 hours

Course type: elective

Languages: slovene and english

Learning and teaching methods:
• lectures • tutorials • students’ individual work