compute_likelihood: Likelihood plot of a two parameter model

In jmzobitz/demodelr: Simulating Differential Equations with Data

Likelihood plot of a two parameter model

Description

compute_likelihood computes the likelihood for a model

Usage

compute_likelihood(model, data, parameters, logLikely = FALSE)

Arguments

model	a function or model of our situation, written with formula notation
data	Data frame of data First column is the independent variable, second column dependent variable. Must be a data.frame
parameters	The data frame matrix of values of the parameters we are using. This will be made using expand.grid or equivalent
logLikely	Do we compute the log likelihood function (default is FALSE). NOTE: what gets returned is - logLikely - meaning that this will be a positive number to work with.

Value

A list with two entries: (1) the likelihood values and (2) values of parameters that optimize the likelihood.

Examples

### Contour plot of a logistic model for two parameters K and b
### using data collected from growth of yeast population

# Define the solution to the differential equation with
# parameters K and b Gause model equation
gause_model <- volume ~ K / (1 + exp(log(K / 0.45 - 1) - b * time))
# Identify the ranges of the parameters that we wish to investigate
kParam <- seq(5, 20, length.out = 100)
bParam <- seq(0, 1, length.out = 100)
# Allow for all the possible combinations of parameters
gause_parameters <- expand.grid(K = kParam, b = bParam)
# Now compute the likelihood
gause_likelihood <- compute_likelihood( model = gause_model,
                                       data = yeast,
                                       parameters = gause_parameters,
                                       logLikely = FALSE
)

jmzobitz/demodelr documentation built on March 6, 2024, 8:31 p.m.