In MapSentinel/sraster:

knitr::opts_chunk$set(echo = TRUE)

Accuracies per class

library(ggplot2)
file_comparision_models = 'C:\\IPSTERS\\sraster\\Results\\Accuraccies_perclass.csv'

#reading file
comparision_models  = read.csv(file_comparision_models, sep = ";",header = T, dec = ',')

comparision_models$Class = factor(comparision_models$Class,levels = c("urban","baresoil", "rainfed", "irrigated", "rice field", "A_grassland", "broadleaf", "conifers", "N_grassland", "shrubland", "S_vegetation", "wetland", "water"))

comparision_models$External= factor(comparision_models$External , levels = c('Baseline','COS','COS_HRL','COS_GL30_HRL'))

comparision_models$Samples <- as.factor(comparision_models$Samples)

comparision_models$Polygons <- as.factor(comparision_models$Polygons)

comparision_models1 = comparision_models[comparision_models$External %in% c('Baseline','COS'),]

#ggplot
p1 <- ggplot(data = comparision_models1 , aes(x=Class, y=F1, fill = External )) + geom_boxplot() + 
  theme(axis.text.x = element_text(size=15, 
                                    face="bold.italic",angle = 90),
        axis.text.y = element_text(angle = 0,size=12,
                                      face="bold.italic"),
        axis.title.y = element_text(size=14, face="bold"),
        axis.title.x = element_text(size=14, face="bold")) +
        expand_limits(y = c(0, 1)) + scale_fill_manual(values=c("#C4961A", "#FC4E07"))
print(p1)

The following graphic shows the

comparision_models2 = comparision_models[comparision_models$External %in% c('COS','COS_HRL','COS_GL30_HRL'),]

#ggplot
p1 <- ggplot(data = comparision_models2 , aes(x=Class, y=F1, fill = External )) + geom_boxplot() + 
  theme(axis.text.x = element_text(size=15, 
                                    face="bold.italic",angle = 90),
        axis.text.y = element_text(angle = 0,size=12,
                                      face="bold.italic"),
        axis.title.y = element_text(size=14, face="bold"),
        axis.title.x = element_text(size=14, face="bold")) +
        expand_limits(y = c(0, 1)) +scale_fill_manual(values=c("#FC4E07","#0072B2","#009E73"))
print(p1)

library(ggplot2)
library(reshape2)

path_file = 'C:\\IPSTERS\\sraster\\ins4\\sampling_4\\output_broadleaf2_Majorityrule_bdist.csv'
data_class = read.csv(path_file, sep= ',', header = T, dec = '.' )

col_ndvi = grep("NDVI",colnames(data_class))
data_class_ndvi = data_class[,c(3,col_ndvi)]

func_stat <- function(l){quantile(l,c(0.5),na.rm=TRUE)}
result_aggregate_time = aggregate(data_class_ndvi, by = list(data_class_ndvi$Object), FUN = func_stat)
resultmelt <- melt(result_aggregate_time[,-1], id.vars = "Object")
#removing ndvi text
resultmelt$variable <- as.Date(substr(resultmelt[,c("variable")], 2, 11),"%Y.%m.%d")
resultmelt$Object <- as.factor(resultmelt$Object)

p1 <- ggplot(resultmelt, aes(variable, value, group = Object)) + 
  geom_line(color="red") + theme(legend.position="top")
print(p1)

MapSentinel/sraster documentation built on April 24, 2020, 5:43 a.m.