Data and Text Mining

professor: prof. dr. Nada Lavrač

professor: doc. dr. Petra Kralj Novak

professor: prof. dr. Bojan Cestnik

professor: prof. dr. Dunja Mladenić

Course

• Information and Communication Technologies (ICT2): Data and Text Mining / Podatkovno in tekstovno rudarjenje (20 ECTS)

Scheduled lectures and course materials in school year 2020/2021

The lectures will be held by Zoom video conference.

The course is divided into three parts:

• data preprocessing (lectured by prof. dr. Bojan Cestnik),
• data mining (lectured by prof. dr. Nada Lavrač and doc. dr. Petra Kralj Novak),
• text mining (lectured by prof. dr. Dunja Mladenić).

Materials:

• Nada Lavrac: Lecture notes
• Petra Kralj Novak:
  
  --- Nov. 11, 2020 ---
  • Lecture notes from Nov. 11, 2020
  • Decision trees (.pdf)
  • Entropy and information gain (.pdf)
  • Hands on: Orange workflow on language bias of decision trees (.ows) and data (.csv)
  
  --- Nov. 25, 2019 ---
  • Lecture notes from Nov. 25, 2020

Video lectures

• Seminar: AI for Industry and Society, Ljubljana 2020
• Orange tutorials

Practice materials: dr. Petra Kralj Novak

• Hands on Scikit-learn and Keras including data gitlab repository

Literature:

1. Bramer, Max: Principles of Data Mining (2007)
2. J. Fuernkranz, D. Gamberger and N. Lavrač, Book: Foundations of Rule Learning, Chapter: Machine Learning and Data Mining, Springer, 2012
3. Loh, Wei‐Yin: Classification and regression trees. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery 1.1 (2011): 14-23.
4. Fawcett, Tom. "An introduction to ROC analysis." Pattern recognition letters 27.8 (2006): 861-874.
5. Aggarwal, Charu C. Data mining: the textbook . Springer, 2015. Chapter 6 : cluster analysis, pgs 195 and 201
6. Nielsen, Michael: Neural Networks and Deep Learning, Chapter 1: Using neural nets to recognize handwritten digits

Course requirements, data mining part:

• Attending lectures, practical exercises and theory exercises
• Oral exam
• Seminar: half a page proposal, 4 pages written report, oral presentation, all in English

1. Data mining seminar
   • Half a page seminar proposal
   • Data analysis of your own data (Orange or scikit-learn recommended)
   • Deliver a 4 pages written report (printed and electronic copy) in Information Society paper format
     on seminar presentations day (use paper template and guidelines)
   • Oral presentation of seminar results (10 minutes for presentation + 5 minutes discussion, use slides template)

Examples of data mining seminars:

• Janez Bucik (.pdf)
  Microsoft stock quotes dependency analysis
• Valentin Koblar (in Slovene .pdf)
  Napoved menjalnega tečaja ameriškega dolarja na podlagi menjalnih tečajev tujih valut

Ideas for seminars

1. Analyze some data where you are the domain expert, use at least two algorithms
2. Find some interesting data to analyze, possible sources:
   • Statistični urad Republike Slovenije
   • Ljubljanska borza d.d.
   • Banka Slovenije
   • World Health Organization
   • Center for Climatic Research, University of Delaware

Templates

• presentation template (.pot)
• paper template (.doc)
• paper guidelines (.doc)

Useful links

• Orange, datasets for practice and tutorials
• Scikit-learn
• Weka
• ClowdFlows

Link to last year's web page - Data and Text Mining 2019/2020
Last update: January 12, 2021