R/bt_estimate_detective.R

In global-trade-alert/gtabastiat:

# Roxygen documentation

#' Bastiat, please re-estimate this detective.
#'
#' @return Several data frames about how the detective did.
#' @references www.globaltradealert.org
#' @author Johannes Fritz for GTA


# Function infos and parameters  --------------------------------------------

bt_estimate_detective = function(detective.name=NULL,
                                 robustness.turns=0,
                                 train.share=.82,
                                 save.stats=T,
                                 save.classifier=T,
                                 training.data.path="data/classifier/training data.Rdata",
                                 training.data.update=F
                                 ){


  #for testing
  # robustness.turns=0
  # train.share=.82
  # save.stats=T
  # save.classifier=T
  # training.data.path="data/classifier/training data.Rdata"
  # training.data.update=F



  if(training.data.update){
    bt_update_training_data()
  }

  load(training.data.path)

  #testing
  #detective = detective.name[1]

  for(detective in detective.name){
    print(paste("Starting",detective))


    #for testing
    # bid=training$bid
    # evaluation=training$evaluation
    # text=training$text
    # acting.agency = training$acting.agency
    # train.share = train.share
    # detective.name=detective

    detective.data=bt_create_estimation_data(bid=training$bid,
                                            evaluation=training$evaluation,
                                            text=training$text,
                                            acting.agency = training$acting.agency,
                                            train.share = train.share,
                                            detective.name=detective
                                          )



    print("Prepared data")

    #dbg

    # training.data=detective.data$estimation.data
    # training.id="bid"
    # training.y="evaluation"
    # estimation.model=detective.data$detective.characteristics$estimation.model
    # robustness.turns=robustness.turns
    # train.share=train.share
    # update.log=save.stats
    # detective.name=detective.data$detective.characteristics$detective.name

    detective.classifier=bt_estimate_classifier(training.data=detective.data$estimation.data,
                                             training.id="bid",
                                             training.y="evaluation",
                                             estimation.model=detective.data$detective.characteristics$estimation.model,
                                             robustness.turns=robustness.turns,
                                             train.share=train.share,
                                             update.log=save.stats,
                                             detective.name=detective.data$detective.characteristics$detective.name,
                                             max.training.rows = 10000)


    if(save.classifier){

      classifier=detective.classifier$classifier
      cutoff=detective.classifier$cutoff
      word.score=detective.data$word.score

      save(classifier, cutoff, word.score, file=paste("content/0 core/",Sys.Date()," - ",detective.data$detective.characteristics$detective.name," classifier.Rdata", sep=""))

      if(detective=="incumbent"){

        save(classifier, cutoff, word.score, file="content/0 core/Bastiat classifier.Rdata")

      }
    }

    print(paste("Finish",detective))
    print(paste("Score: ",detective.classifier$performance$score,"; Score adjusted: ",detective.classifier$performance$score.adjusted, sep=""))
  }
}