Articles
PREDICTING TIME TO FLOWERING WITH A FIRST PASSAGE TIME MODEL
Article number
456_17
Pages
151 – 160
Language
Abstract
Time to first visible cyathium (VC) and first visible flower (VF) in poinsettias (Euphorbia pulcherrima Willd.) is predicted using a First Passage Time (FPT) model.
The FPT concept is developed for modelling the occurrence of discrete events in a time dependent process.
An FPT model also makes it possible to quantify relevant confidence intervals for the predictions, which is important in practical applications.
The data sets include the potential effect of artificial light intensity, air temperature, temperature drop treatment, plant density and number of vegetative days.
A non-dimensional development index is modelled as a linear combination of their integrals.
A sequential cross-validation procedure is used to quantify the prediction error within the calibration data set of every possible sub model.
From this a model of optimal complexity with respect to prediction can be found.
Comparison of predicted and observed time to VC and VF in independent data sets are used to assess the validity of the calibration data.
The FPT concept is developed for modelling the occurrence of discrete events in a time dependent process.
An FPT model also makes it possible to quantify relevant confidence intervals for the predictions, which is important in practical applications.
The data sets include the potential effect of artificial light intensity, air temperature, temperature drop treatment, plant density and number of vegetative days.
A non-dimensional development index is modelled as a linear combination of their integrals.
A sequential cross-validation procedure is used to quantify the prediction error within the calibration data set of every possible sub model.
From this a model of optimal complexity with respect to prediction can be found.
Comparison of predicted and observed time to VC and VF in independent data sets are used to assess the validity of the calibration data.
The FPT concept relies heavily on numerical solutions of integrals and equations, and extensions of the model towards a more physiologically based structure is therefore straightforward.
Publication
Authors
L. Snipen, R. Moe, J. Søreng
Keywords
Poinsettia, Flower development, Cross-validation
Online Articles (65)