predict.fast_multinom: Prediction method for fast_multinom fits

In MultinomialMutations/MultinomialMutations: Fast multinomial regression for large cancer mutation datasets

Description

Obtains predictions from a multinomial model fitted using fast_multinom.

Usage

## S3 method for class 'fast_multinom'
predict(object, newdata = NULL)

Arguments

object	Object of class fast_multinom.
newdata	Optionally, a data frame for which to predict.

Details

Only output on the level of probabilities are generated. This is equivalent to type="response" in a logistic regression model estimated with glm with family=binomial(link="logit").

Note that the factor levels and contrasts defined in the fast_multinom object are used. This allows to run prediction on a dataset with fewer levels and different contrasts (in this case, the contrasts are overwritten and there is a warning. There can be the same warning, when the contrasts are not changed – I don't know why, it is issued by model.Matrix.)

Author(s)

Johanna Bertl

References

Bertl, J.; Guo, Q.; Rasmussen, M. J.; Besenbacher, S; Nielsen, M. M.; Hornshøj, H.; Pedersen, J. S. & Hobolth, A. A Site Specific Model And Analysis Of The Neutral Somatic Mutation Rate In Whole-Genome Cancer Data. bioRxiv, 2017. doi: https://doi.org/10.1101/122879 http://www.biorxiv.org/content/early/2017/06/21/122879

Examples

data(cancermutations)

# the APOBEC signature is only relevant for transitions and transversions to a G:C basepair -- construct the corresponding subset of parameters for the 3 binomial models:
subs = matrix(T, ncol=3, nrow=4)
subs[3,2] = F

# fit the multinomial model
fit = fast_multinom(cbind(NO, I, VA, VG) ~ strong + apobec + cancer_type, data = cancermutations, refLevel=1, VC=T, subsetmatrix=subs, predictions=T)

# predictions on the data that was used for fitting (only available, because predictions=T in the function fast_multinom):
head(predict.fast_multinom(fit))

# predict on new data (with fewer factor levels):
set.seed(123)
new = droplevels(cancermutations[sample.int(nrow(cancermutations), 5),])
pred = predict.fast_multinom(fit, new)

MultinomialMutations/MultinomialMutations documentation built on May 22, 2019, 4:39 p.m.