mcmc_estimate {demodelr} R Documentation

## Markov Chain parameter estimates

### Description

mcmc_estimate Computes and Markov Chain Monte Carlo parameter estimate for a given model

### Usage

mcmc_estimate(
model,
data,
parameters,
iterations = 1,
knob_flag = FALSE,
mode = "emp",
initial_condition = NULL,
deltaT = NULL,
n_steps = NULL
)


### Arguments

 model the model equations that we use to compute the result. data the data used to assess the model parameters a data frame that lists the names of the parameters along with upper and lower bounds iterations the number of iterations we wish to run the MCMC for. knob_flag determines if we tune the range that can be search (annealing) mode two choices: emp –> empirical (default) or de –> differential equations. The estimator works differently depending on which is used. initial_condition The initial condition for the differential equation (DE mode only) deltaT The length between timesteps (DE mode only) n_steps The number of time steps we run the model (DE mode only)

### Value

A dataframe: the first column is the accept flag of the mcmc run (TRUE/FALSE), the log likelihood, and the parameter values

mcmc_analyze

### Examples



## Example with an empirical model:
## Step 1: Define the model and parameters
phos_model <- daphnia ~ c * algae^(1 / theta)

phos_param <- tibble::tibble( name = c("c", "theta"),
lower_bound = c(0, 1),
upper_bound = c(2, 20))

## Step 2: Determine MCMC settings
# Define the number of iterations
phos_iter <- 1000

## Step 3: Compute MCMC estimate
phos_mcmc <- mcmc_estimate(model = phos_model,
data = phosphorous,
parameters = phos_param,
iterations = phos_iter)

## Example with a differential equation:
## Step 1: Define the model, parameters, and data
## Define the tourism model
tourism_model <- c(dRdt ~ resources * (1 - resources) - a * visitors,
dVdt ~ b * visitors * (resources - visitors))

# Define the parameters that you will use with their bounds
tourism_param <- tibble::tibble( name = c("a", "b"),
lower_bound = c(10, 0),
upper_bound = c(30, 5))

## Step 2: Determine MCMC settings
# Define the initial conditions
tourism_init <- c(resources = 0.995, visitors = 0.00167)
deltaT <- .1 # timestep length
n_steps <- 15 # must be a number greater than 1
# Define the number of iterations
tourism_iter <- 1000

## Step 3: Compute MCMC estimate
tourism_out <- mcmc_estimate(
model = tourism_model,
data = parks,
parameters = tourism_param,
mode = "de",
initial_condition = tourism_init, deltaT = deltaT,
n_steps = n_steps,
iterations = tourism_iter)



[Package demodelr version 1.0.1 Index]