| Semester | Herbstsemester 2017 |
| Angebotsmuster | Jedes Semester |
| Dozierende | Erich Studerus (erich.studerus@unibas.ch, BeurteilerIn) |
| Inhalt | Learning from data and making accurate predictions are essential for any scientific discipline. For example in medicine and psychology, prediction models can help health care providers in estimating the probability or risk that a specific disease or condition is present or that a specific event will occur in the future, which can improve clinical decision making. With the explosion of "Big data" problems and the developement of increasingly sophisticated predictive algorithms, so called machine learning methods, prediction modelling has become a very hot topic in many scientific areas in recent years, including psychology. The aim of this seminar is to give an introduction to prediction modeling and its application in psychology. Participants will learn the basic principles, advantages and disadvantages, as well as practical applications of the most important statistical learning methods, including both traditional (e.g. logistic regression) and more modern machine learning methods (e.g. lasso, random forest, support vector machines). They will learn how to optimally train these models and how to assess their predictive performance using cross validation. Furthermore, they will learn how to deal with highdimensional data problems (small n and large number of variables), missing data, and class imbalance. For practical applications, we will use the feely available machine learning in R (mlr) package. Participants are therefore required to have at least basic skills in using R. Participants who are not yet familiar with R are adviced to complete the free online R tutorials on www.datacamp.com before the start of the seminar. |
| Lernziele | The participants should know how to develop and validate a prediction model using the mlr package in R. They should have a basic understanding of the most important statistical learning methods and know how to deal with fundamental problems in prediction modelling, such as overfitting, large number of potential predictor variables relative to the number of cases and missing data. |
| Literatur | Ewout W. Steyerberg, 2009, Clinical Prediction Models - A Practical Approach to Development, Validation, and Updating, Springer Verlag, James, Witten, Hastie & Tibshirani, 2013, An Introduction to Statistical Learning with Applications in R, Springer Verlag Kuhn, Johnson, 2013, Applied Predictive Modelling, Springer Verlag. |
| Bemerkungen | Maximale Anzahl Studierender: 15. |
| Teilnahmebedingungen | Abgeschlossenes Masterstudium. |
| Anmeldung zur Lehrveranstaltung | Das Anmeldeverfahren für die Seminare ist an das Belegen gekoppelt. Das heisst, es läuft über MOnA. Das Belegen der Seminare ist möglich ab dem 01. August 2017. Anmeldefrist ist der 18. August 2017. Ab dem 21. August werden die Dozierenden die Einteilung zu den Seminaren vornehmen und die Studierenden betreffend Einteilung informieren. |
| Unterrichtssprache | Deutsch |
| Einsatz digitaler Medien | kein spezifischer Einsatz |
| HörerInnen willkommen |
| Intervall | Wochentag | Zeit | Raum |
|---|
Keine Einzeltermine verfügbar, bitte informieren Sie sich direkt bei den Dozierenden.
| Module |
Doktorat Psychologie: Empfehlungen (Promotionsfach: Psychologie) Doktoratsstudium Psychologie: Empfehlungen (Promotionsfach: Psychologie (Studienbeginn vor 01.02.2015)) |
| Leistungsüberprüfung | Lehrveranst.-begleitend |
| An-/Abmeldung zur Leistungsüberprüfung | Anm.: Belegen Lehrveranstaltung; Abm.: stornieren |
| Wiederholungsprüfung | keine Wiederholungsprüfung |
| Skala | Pass / Fail |
| Wiederholtes Belegen | beliebig wiederholbar |
| Zuständige Fakultät | Fakultät für Psychologie, studiendekanat-psychologie@unibas.ch |
| Anbietende Organisationseinheit | Fakultät für Psychologie |