In PabloMBooster/SISESATools: What the package does (short line)

knitr::opts_chunk$set(echo = TRUE)
library(ggplot2)

VMS data analysis

This 'package' (set of codes) is intended for Peruvian VMS (SISESAT) data applications. When I first started using SISESAT data, I was coding in Matlab. In collaboration with Pablo Marin, we rewrote a great part of those codes in R; particularly the ones for data pre-processing. If you want to cite any of this, you should refer to https://github.com/PabloMBooster/SISESATools. During all these years, the data given to us has taken different formats, each one with its specific problems. Here, we will not deal with those specific problems. Instead, we will start with a given format and its up to the users to adapt the codes to their data or their datasets to fit the codes.

The codes will comprise the following stages:

\begin{itemize} \item Pre-processing: some codes to clean the data. \item Training dataset: crossing VMS and observers data. Use only when observers data or reliable logbook data is available. \item ANN: training and validating artificial neural networks to identify fishing set locations in the data; then use it for prediction \end{itemize}

Some other codes could be added in the future.

Pre-processing

Let's start by indicating R where the files are and will be.

# setwd("/run/media/rjoo/data/VMScripts/Scripts/SISESATools/")
dossier.data.inputs <- './data/' # that should exist and that's where you stock data
dossier.functions <- './R/' # that's where you stock Scripts
dossier.data.outputs <- './DataOutputs/' # that's where you'll put data outputs later
dossier.stat.outputs <- './StatOutputs/' # that too
#checking if the last two exist qnd if not, create
if (!dir.exists(dossier.data.outputs)){
  dir.create(dossier.data.outputs)
}
if (!dir.exists(dossier.stat.outputs)){
  dir.create(dossier.stat.outputs)
}

# I have to transform a data to make a playing dataset
library(random)
library(randomNames)
source(paste0(dossier.functions,'AnonymData.R'))
data <- read.table(file = paste0(dossier.data.inputs,'RealData.txt'),header = TRUE,sep="\t",dec = ".")
# just take the first 100000 lines
num.obs <- 100000
data.modif <- anonym.data(data,num.obs)

save(data.modif,file=paste0(dossier.data.inputs,'play.RData'))

write.table(data.modif,file=paste0(dossier.data.inputs,"VMSdataset.txt"),sep="\t",dec = ".",row.names = FALSE,col.names=TRUE)

# Directory of data base
VMSdataset <- system.file("extdata", "VMSdataset.txt", package = "SISESATools")
print(VMSdataset)

# Read a example of data base
VMSdataset <- read.table(VMSdataset, header = T)#, sep="\t", dec = ".")
head(VMSdataset)

#source(paste0(dossier.functions,"AuxiliaryFunctions.R")) # tiene lo mismo o menos que la de Pablo
#source(paste0(dossier.functions,"sisesat-main.R")) 
#source(paste0(dossier.functions,"sisesat-internal.R")) 

# cleaning data and computing first variables