Package 'likelihoodTools'

Extract and format results from Simulated Annealing (Maximum Likelihood Estimation)

Description

Extract and format results from Simulated Annealing (Maximum Likelihood Estimation)

Usage

mle_format(x, yvar)

Arguments

x	List with the results of the simulated annealing algorithm for Maximum Likelihood Estimation. See likelihood::anneal()
yvar	The name of the column that contains the dependent variable (the “observed” value). This column must be present in the source_data of the x (results) list

Value

A dataframe with outputs from the results of the simulated annealing maximum parameter estimation. This dataframe contains the following columns (see help in likelihood::anneal()):

• max_likeli The maximum likelihood value of the model

• n_params The number of the estimated parameters

• aic_corr The value of Akaike's Information Criterion “corrected” for small sample size. See the "Simulated Annealing Algorithm" help page of the likelihood package for more information.

• aic The value of Akaike's Information Criterion. See the "Simulated Annealing Algorithm" help page of the likelihood package for more information. slope Slope of observed values linearly regressed on those predicted by model, using the parameter maximum likelihood estimates. The intercept is forced at zero.

• R2 Proportion of variance explained by the model relative to that explained by the simple mean of the data.

• rmse Root Mean Square Error, i.e. the standard deviation of the residuals. It is computed as:

$RMSE=\sqrt{ \frac{\sum_{i=1}^{N}(obs_i - exp_i)^2}{n -1}}$

Examples

# Get the results of the maximum likelihood estimation from the example in
# the anneal function of the likelihood pkg.

library(likelihood)
data(crown_rad)
dataset <- crown_rad

# Create our model function
modelfun <- function (a, b, DBH) {a + b * DBH}

# Compute the MLE of the parameters
results <- anneal(model = modelfun,
  par = list(a = 0, b = 0),
  var = list(DBH = "DBH", x = "Radius", mean = "predicted",
             sd = 0.815585, log = TRUE),
  source_data = dataset,
  par_lo = list(a = 0, b = 0),
  par_hi = list(a = 50, b = 50),
  pdf = dnorm,
  dep_var = "Radius",
  max_iter = 20000,
  show_display = FALSE)

# Format the results
mle_format(results, yvar = "DBH")

---

Plots Observed vs. Predicted MLE

Description

Plots observed values vs. predicted values. The predicted values are obtained from the model with the parameters values estimated by maximum likelihood estimation using simulated annealing.

Usage

mle_plot_observed(
  x,
  yvar,
  annotate = TRUE,
  lab_x = "Observed",
  lab_y = "Predicted",
  ...
)

Arguments

x	List with the results of the simulated annealing algorithm for Maximum Likelihood Estimation. See likelihood::anneal()
yvar	The name of the column that contains the dependent variable (the “observed” value). This column must be present in the source_data of the x (results) list
annotate	logical (default to TRUE), display the values of $R^2$ and slope of the regression of the observed on predicted values. See likelihood::Simulated Annealing Algorithm
lab_x	The text for the x-axis lab
lab_y	The text for the y-axis lab
...	other ggplot2 parameters

Value

A ggplot object displaying the observed vs. predicted values, with optional annotations for $R^2$ and regression slope.

---

Plots Residuals vs. Predicted MLE

Description

Plots residuals (observed - residuals) values vs. predicted values. The predicted values are obtained from the model with the parameters values estimated by maximum likelihood estimation using simulated annealing.

Usage

mle_plot_residuals(
  x,
  yvar,
  lab_residuals = "Residuals",
  lab_predicted = "Predicted",
  ...
)

Arguments

x	List with the results of the simulated annealing algorithm for Maximum Likelihood Estimation. See likelihood::anneal()
yvar	The name of the column that contains the dependent variable (the “observed” value). This column must be present in the source_data of the x (results) list
lab_residuals	The text for the residual axis lab (y-axis)
lab_predicted	The text for the predicted axis lab (x-axis)
...	other ggplot2 parameters

Value

A ggplot object displaying the residuals vs. predicted values, with a horizontal line at zero.

---