Leave-one-out cross-validation (abbreviated as loo-cv or cv-loo) is an internal cross validation technique (for futher details please see ref.1). loo-cv is the first (m = 1) in the series of leave-many-out cross-validation (lmo-cv) and it is a alternative with 100% results reproducibility to k-fold cross-validation. Employ n training sets and from each of these one compound is excluded. For each training set a model is obtained and then it is used to predict the property/activity of excluded compound. The program apply leave-one-out algorithm for predicted dependent variable. Please input (or copy/paste) data in Textbox, with 'space' or 'tab' separator between columns and 'enter' separator between rows as in sample data.
Ref.:
1. Bolboaca, S.D.; Jäntschi, L. Modelling the property of compounds from structure: statistical methods for models validation. Environmental Chemistry Letters 2008, 6(3), 175 - 181. DOI 10.1007/s10311-007-0119-9