#dane z http://neuroph.sourceforge.net/tutorials/ForestFires/PredictingTheBurnedAreaOfForestFires.html
#skróty z nazw tabel http://cwfis.cfs.nrcan.gc.ca/background/summary/fwi

#czyścimy
gc(reset=T)
rm(list = ls())

#ładujemy potrzebne biblioteki
library ('dplyr')
#pobieranie pliku
path <-"D:/forestfire.csv"
dane<-read.csv(path,header=TRUE,sep=";",dec=',')
dane%>%subset(area<=40)%>%subset(area>0)->dane2

FindOutliers <- function(data) {
  lowerq = quantile(data)[2]
  upperq = quantile(data)[4]
  iqr = upperq - lowerq #Or use IQR(data)
  # we identify extreme outliers
  extreme.threshold.upper = (iqr * 3) + upperq
  extreme.threshold.lower = lowerq - (iqr * 3)
  result <- which(data > extreme.threshold.upper | data < extreme.threshold.lower)
}

temp <- FindOutliers(dane2$FFMC)
dane2$FFMC[temp]<-NA

#histogram
hist(dane2$area
     ,col= 'green'#kolor histogramu
     ,main='histogram obszaru spalonego'#nazwa histogramu
     ,xlab='area'#nazwa x-owej
     ,ylab='liczba wystapien')#nazwa y-kowej

#korelacja
dane2%>%select_("area", "FFMC","DMC","DC","ISI"  ,         
               "temp","RH","wind","rain")->dane_cov
CORELATIONS <- cor(dane_cov, method="pearson", use="complete.obs")
round(CORELATIONS,3)->CORELATIONS
library(corrplot)
corrplot(CORELATIONS, type="lower", tl.col="black", tl.srt=45)

#wkresy zależności
plot(dane2$area, dane2$temp,
     col="black",pch=21,bg="red",cex=2,ylab="",xlab="",
     axes=TRUE,main="Zaleznosc obszaru spalonego od temperatury")
mtext('area', side=1,line=2.5,cex=1.5)
mtext('temp', side=2,line=2.5,cex=1.5)
abline(linear_model, lty=5,lwd=2, col="blue")

plot(dane2$area, dane2$FFMC,
     col="black",pch=21,bg="red",cex=2,ylab="",xlab="",
     axes=TRUE,main="Zaleznosc obszaru spalonego od indexu ffmc")
mtext('area', side=1,line=2.5,cex=1.5)
mtext('FFMC', side=2,line=2.5,cex=1.5)

plot(dane2$area, dane2$DMC,
     col="black",pch=21,bg="red",cex=2,ylab="",xlab="",
     axes=TRUE,main="Zaleznosc area od DMC")
mtext('area', side=1,line=2.5,cex=1.5)
mtext('RH', side=2,line=2.5,cex=1.5)

plot(dane2$ISI, dane2$FFMC,
     col="black",pch=21,bg="red",cex=2,ylab="",xlab="",
     axes=TRUE,main="Zaleznosc isi od indexu ffmc")
mtext('ISI', side=1,line=2.5,cex=1.5)
mtext('FFMC', side=2,line=2.5,cex=1.5)

plot(dane2$temp, dane2$RH,
     col="black",pch=21,bg="red",cex=2,ylab="",xlab="",
     axes=TRUE,main="Zaleznosc temp od rh")
mtext('temp', side=1,line=2.5,cex=1.5)
mtext('RH', side=2,line=2.5,cex=1.5)

#regresja liniowa
fit1 = lm(formula = dane2$area ~ dane2$temp)
summary(fit1)
plot(dane2$area,predict(fit1))
abline(fit1)

#regresja wielowymiarowa