sits_train: Train SITS classification models
In luizassis/sits: Satellite Image Time Series Analysis

Description Usage Arguments Details Value Author(s)

Given a tibble with a set of distance measures, returns trained models using support vector machines. This function will use the TWDTW alignment information for all classes as the attributes of the chosen machine learning methods. Please use this function in the following way: 1. call sits_patterns to produce a set a labelled patterns from a reference data set 2. call a method to get distances between a time series and patterns to produce a set of alignements 3. use the distances tibble as an input to the training function

1	sits_train(distances.tb, tr_method = sits_svm())

`distances.tb`	a time series with a set of distance measures for each training sample
`tr_method`	a traning method that returns a model for prediction

Functions for machine learning associated to the SITS package The attributes for the training functions are the DTW distances computed by the TWTDW function (see documentation on sits_TWDTW_matches)

models supported: 'svm', 'random forests', 'boosting', 'lda', 'qda' 'multinomial logit', 'lasso', 'ridge', 'elnet', 'best model'

result a model fitted into input data given by train_method parameter

Rolf Simoes, rolf.simoes@inpe.br

Alexandre Xavier Ywata de Carvalho, alexandre.ywata@ipea.gov.br

luizassis/sits documentation built on May 30, 2019, 7:15 p.m.