Type: Package
Title: Tabular Maximum Likelihood Classifier
Version: 0.0.3
Description: The maximum likelihood classifier (MLC) is one of the most common classifiers used for remote sensing imagery. This package uses 'RcppArmadillo' to provide a fast implementation of the MLC to train and predict over tabular data (data.frame). The algorithms were based on Mather (1985) <doi:10.1080/01431168508948456> method.
License: GPL-3
Depends: Rcpp, methods
Imports: stats
LinkingTo: Rcpp, RcppArmadillo
RoxygenNote: 7.1.2
Encoding: UTF-8
URL: https://github.com/caiohamamura/tabularMLC
BugReports: https://github.com/caiohamamura/tabularMLC/issues
Author: Caio Hamamura ORCID iD [aut, cre]
Maintainer: Caio Hamamura <caiohamamura@gmail.com>
Repository: CRAN
Repository/R-Forge/Project: tabularmlc
Repository/R-Forge/Revision: 5
Repository/R-Forge/DateTimeStamp: 2021-10-04 20:39:29
Date/Publication: 2021-10-05 08:30:02 UTC
NeedsCompilation: yes
Packaged: 2021-10-04 20:48:40 UTC; rforge

Tabular maximum likelihood classifier

Description

Maximum likelihood is a common classifier used for land use classification. It calculates the likelihood of an object to belong to each class based on an expected distribution and a metric of distance.

Details

The most common implementation, like in this package, will assume normal distributed variables within classes, and calculate the distance, based on Mahalanobis distance.

Author(s)

Maintainer: Caio Hamamura caiohamamura@gmail.com (ORCID)

References

Mather, P. M. (1985). Remote sensing letters: A computationally efficient maximum-likelihood classifier employing prior probabilities for remotely-sensed data. International Journal of Remote Sensing, 6(2), 369–376. doi: 10.1080/01431168508948456

Imports

See Also

Useful links:

Maximum Likelihood Classifier

Description

Function to create the classifier class from the training set

Usage

MLC(x, ...)

## S3 method for class 'formula'
MLC(formula, data = NULL, ...)

## Default S3 method:
MLC(x, y = NULL, ...)

Arguments

x

feature vector for the training set

...

for other signatures

formula

formula. The formula for defining the model.

data

the dataset

y

factor vector with the training set labels

Value

An object of class MLC.model parameters used for the model

Examples

data(iris)

x = iris[, -5]
y = iris$Species

# Default x y interface
mlcModel1 = MLC(x, y)

# Formula interface
mlcModel2 = MLC(Species ~ Petal.Length + Petal.Width, iris)

# Formula except one column
mlcModel3 = MLC(Species ~ . - Sepal.Length, iris)

Maximum likelihood model class

Description

Maximum likelihood model class

Slots

k

the constant fraction to be used in model \frac{1}{(2 \pi)^{\frac{L}{2}} \sqrt{\left | \Sigma_i \right |}}

mu

mean (\mu_i) list for each variable and class

inverseCovarianceMatrices

inverted covariance matrix (\Sigma_i) for each class

groups

the classification levels

vars

the variables used for trainning the model

See Also

MLC which creates this class

Predict function for MLC.model-class

Description

predict is inherited from the generic function for predictions from the results.

Usage

## S3 method for class 'MLC.model'
predict(object, x = NULL, likelihood = FALSE, ...)

Arguments

object

MLC.model-class model class to use for prediction

x

data.frame. The feature vector to predict

likelihood

logical. Whether to return or not the likelihood values, default FALSE.

...

inherited from generic function (not in use)

Value

a factor vector with the predicted value. If likelihood is TRUE, then it will also return the calculated likelihoods.

Examples

data(iris)

n = length(iris$Species)

# Split training by sample
training = sample(1:n, size=n*0.7)
validation = (1:n)[-training]

# Train model with training dataset
mlcModel = MLC(Species ~ ., iris[training,])

# Predict using validation dataset
predict = predict(mlcModel, iris[validation,])

# Print confusion matrix
confusionMatrix = table(predicted=predict, observed=iris$Species[validation])
print(confusionMatrix)