add_observation-methods: Add an observation

In MKaptein/ofmlr: Fit online finite mixtures of logistic regression models

Description

Generic function for adding observations.

The function takes as first argument an initialized model (or a collection of models), and subsequently updates the parameters given an observation that is split into y (a 0 or 1 scalar) and a vector X containing the features. Note that the lenght of X needs to match the dimensions of the parameters of the current model.

The function takes as first argument an initialized model comparison, and subsequently update the parameters of each model give an observation that is split into y (a 0 or 1 scalar) and a vector X containing the features. Note that the lenght of X needs to match the dimensions of the parameters of the current model.

Usage

add_observation(object, y, X, ...)

## S4 method for signature 'online_log_mixture'
add_observation(object, y, X, lambda = 0.1)

## S4 method for signature 'multi_online_log_mixture'
add_observation(object, y, X, ...)

Arguments

object	An object of type online_log_mixture
y	A scalar with value 0 or 1; the dependent variable
X	The feature vector of the current observation
lambda	The current learn rate. If <=0 than lambda <- n+1000^(-.5) where n is the number of observations
object	An object of type online_log_mixture
y	A scalar with value 0 or 1; the dependent variable
X	The feature vector of the current observation
lambda	The current learn rate. If <=0 than lambda <- n+1000^(-.5) where n is the number of observations

Value

An updated object of type online_log_mixture

An updated object of type multi_online_log_mixture

Examples

M1 <- online_log_mixture(2,4, trace=1)
M1 <- add_observation(M1, 1, c(2,-3))
M1

M2 <- online_log_mixture(2,1, trace=1)
for(i in 1:500){
		X <- runif(2,-2,2)
		y <- rbinom(1, 1, inv_logit(c(2,-2)%*%X))
		M2 <- add_observation(M2, y, X, .1)
}
M2

M1 <- online_log_mixture(2,1)
models <- multi_online_log_mixture(M1)
models <- add_model(models, online_log_mixture(2,2))
models <- add_observation(models, 1, c(2,-3))

MKaptein/ofmlr documentation built on Sept. 25, 2020, 11:46 a.m.