sits (version 0.13.0)

sits_accuracy: Area-weighted classification accuracy assessment

Description

This function calculates the accuracy of the classification result. For a set of time series, it creates a confusion matrix and then calculates the resulting statistics using the R packge "caret". The time series needs to be classified using sits_classify.

Classified images are generated using sits_classify followed by sits_label_classification. For a classified image, the function uses an area-weighted technique proposed by Olofsson et al. according to [1-3] to produce more reliable accuracy estimates at 95

In both cases, it provides an accuracy assessment of the classified, including Overall Accuracy, Kappa, User's Accuracy, Producer's Accuracy and error matrix (confusion matrix)

Usage

sits_accuracy(data, ...)
# S3 method for sits
sits_accuracy(data, ...)
# S3 method for classified_image
sits_accuracy(data, ..., validation_csv)

Arguments

data

Either a data cube with classified images or a set of time series

…

Specific parameters

validation_csv

A CSV file path with validation data

Value

A list of lists: The error_matrix, the class_areas, the unbiased estimated areas, the standard error areas, confidence interval 95 and the accuracy (user, producer, and overall), or NULL if the data is empty. A confusion matrix assessment produced by the caret package.

References

[1] Olofsson, P., Foody, G.M., Stehman, S.V., Woodcock, C.E. (2013). Making better use of accuracy data in land change studies: Estimating accuracy and area and quantifying uncertainty using stratified estimation. Remote Sensing of Environment, 129, pp.122-131.

[2] Olofsson, P., Foody G.M., Herold M., Stehman, S.V., Woodcock, C.E., Wulder, M.A. (2014) Good practices for estimating area and assessing accuracy of land change. Remote Sensing of Environment, 148, pp. 42-57.

[3] FAO, Map Accuracy Assessment and Area Estimation: A Practical Guide. National forest monitoring assessment working paper No.46/E, 2016.

Examples

# NOT RUN {
# Case (1) - Accuracy for classification of time series
# read a tibble with 400 time series of Cerrado and 346 of Pasture
data(cerrado_2classes)
# create a model for classification of time series
svm_model <- sits_train(cerrado_2classes, sits_svm())
# classify the time series
predicted <- sits_classify(cerrado_2classes, svm_model)
# calculate the classification accuracy
acc <- sits_accuracy(predicted)

# Case (2) - Accuracy for classification of raster data
# select a training set with two bands
samples_modis_2bands <- sits_select(samples_modis_4bands,
                                    bands = c("NDVI", "EVI"))

# build an extreme gradient boosting model
xgb_model <- sits_train(
  samples_modis_2bands,
  sits_xgboost(nrounds = 50, verbose = FALSE)
)

# create a data cube based on files
data_dir <- system.file("extdata/raster/mod13q1", package = "sits")
cube <- sits_cube(
    source = "LOCAL",
    name = "sinop-2014",
    satellite = "TERRA",
    sensor = "MODIS",
    data_dir = data_dir,
    delim = "_",
    parse_info = c("X1", "X2", "tile", "band", "date")
)


# classify the data cube with xgb model
probs_cube <- sits_classify(cube,
  xgb_model,
  output_dir = tempdir(),
  memsize = 4,
  multicores = 2
)
# label the classification
label_cube <- sits_label_classification(probs_cube,
  output_dir = tempdir()
)
# get ground truth points
ground_truth <- system.file("extdata/samples/samples_sinop_crop.csv",
  package = "sits"
)
# calculate accuracy according to Olofsson's method
accuracy <- suppressWarnings(sits_accuracy(label_cube,
            validation_csv = ground_truth))
# }