Zu dieser Karteikarte gibt es einen kompletten Satz an Karteikarten. Kostenlos!
75
Types of cross-validation?
- k-fold Split the data into k parts. Use k-1 parts for Training and the last for validation. Repeat k times.
- 2-fold Split the data into 2 parts. Use each part once for training and validate with the other.
- Leave-one-out Split the data into as many parts as you have data points. Train with all but one and validate with the one left out.
- Repeated random sub-sampling Split the dataset randomly k times. Choose independantly how large each validation set is.
Tags:
Quelle: CI Teil 1 Lecture 6
Quelle: CI Teil 1 Lecture 6
Karteninfo:
Autor: Sepp Samuel
Oberthema: Telematik
Thema: Computational Intelligence
Schule / Uni: TU Graz
Veröffentlicht: 02.07.2014