MJacobs1985

arc_month%>%
  filter(!is.na(sampledate))%>%
  plot_acf_diagnostics(sampledate, rochef, 
                                 .lags=24,
                                 .facet_ncol = 2, 
                                 .facet_scales = "free")
arc_month%>%
  filter(!is.na(sampledate))%>%
  plot_seasonal_diagnostics(sampledate, rochef, .interactive = FALSE)
arc_month %>%

MJacobs1985

arc_month%>%
  filter(!is.na(sampledate))%>%
  plot_anomaly_diagnostics(sampledate, 
                           rochef, 
                           .facet_ncol = 2, 
                           .facet_scales = "free")

MJacobs1985

arc_month<-arc%>%
  summarise_by_time(sampledate,
                    .by="month",
                    Weight=median(wt.s,na.rm=TRUE),
                    rochef=median(rochefant, na.rm=TRUE),
                    nirala=sum(`nir-ala`, na.rm=TRUE),
                    nirdha=median(`nir-dha`, na.rm=TRUE),
                    nirdpa=sum(`nir-dpa`, na.rm=TRUE),
                    nirepa=median(`nir-epa`,na.rm=TRUE),
                    nireta=sum(`nir-eta`,na.rm=TRUE),

MJacobs1985

ggplot(arc, aes(x=rochefant, y=length, colour=(as.factor(generation1)))) + 
  geom_point(alpha=0.4) +
  facet_wrap(~`generation(a/s)`) +
  theme_bw() 
ggplot(arc, aes(x=rochefant, y=length, colour=(as.factor(generation1)))) + 
  geom_point(alpha=0.4) +
  facet_wrap(~weightclass) +
  theme_bw() 
ggplot(arc, aes(x=rochefant, y=length, colour=weightclasshalv)) + 
  geom_point(alpha=0.4) +

MJacobs1985

DataExplorer::plot_missing(arc)
DataExplorer::plot_correlation(arc_month,  
                               cor_args = list("use" = "pairwise.complete.obs"))
DataExplorer::plot_correlation(arc_company,  
                               cor_args = list("use" = "pairwise.complete.obs"))

ggplot(arc, aes(x=rochefant, colour=description)) + 
  geom_density(aes(fill=description), alpha=0.2) + 
  theme_bw() + facet_wrap(~description)
ggplot(arc, aes(x=rochefant, colour=strain)) + 

MJacobs1985

arc<-as.data.frame(arc)
arcraw<-as.data.frame(arcraw)
varPlot(form=rochefant~(region), Data=arc)
varPlot(form=rochefant~(customer), Data=arc) # some customers for sure provided more data then others 
varPlot(form=rochefant~(locality), Data=arc)
varPlot(form=rochefant~(slaughteryear), Data=arc)
varPlot(rochefant~(slaughteryear+slaughtermonth), arc, keep.order = FALSE,
        MeanLine=list(var=c("int", "slaughteryear"), 
                      col=c("magenta", "blue"), lwd=c(2,2)))
varPlot(form=rochefant~(strain), Data=arc)

MJacobs1985

interceptonlymodel <- lmer(formula = rochefant ~ 1 + (1|customer),
                           data    = arc);summary(interceptonlymodel)
interceptonlymodel <- lmer(formula = rochefant ~ 1 + (1|locality),
                           data    = arc);summary(interceptonlymodel)
interceptonlymodel <- lmer(formula = rochefant ~ 1 + (1|region),
                           data    = arc);summary(interceptonlymodel)
interceptonlymodel <- lmer(formula = rochefant ~ 1 + (1|slaughteryear),
                           data    = arc);summary(interceptonlymodel)
interceptonlymodel <- lmer(formula = rochefant ~ 1 + (1|description),
                           data    = arc);summary(interceptonlymodel)

MJacobs1985

## Create summary time-series datasets for quick relationship building
length(levels(arc$customer))
length(levels(arc$locality))
length(levels(arc$region))
length(levels(arc$strain))
length(levels(arc$generation1))
length(levels(arc$`generation(a/s)`))
arc_month<-arc%>%
  summarise_by_time(sampledate,
                    .by="month",

MJacobs1985

### Recode all character to factor data ## 
arc[sapply(arc, is.character)]<-lapply(arc[sapply(arc, is.character)],as.factor)
str(arc)
## Recode Generation (A/S)
levels(arc$`generation(a/s)`)[levels(arc$`generation(a/s)`)=="05SB"]<-NA
### LOOK AT RAW DATA
dim(arc)
str(arc)
skim(arc)
## tabulate the data ###

MJacobs1985

rm(list = ls())
library(readxl)
library(caret)
library(dplyr)
library(readr)
library(DataExplorer)
library(skimr)
library(forecast)
library(lubridate)
library(xts)