omartinez182

x<-seq(1,20)
y<-c(2.8, 2.1, 4, 4.5, 3.8, 3.2, 4.8, 5.4, 4, 3.6, 5.5, 5.8, 4.3, 3.9, 6, 6.4, NA, NA, NA, NA)

data <- data.frame(t=x, sales=y)
data

omartinez182

data %>%
  ggplot(aes(x=t, y=sales))+
  geom_line(col = 'light blue')+
  geom_point()+
  xlab('Time')+
  ylab('Sales')+
  theme_minimal()

omartinez182

ggplot(data, aes(x=t, y=sales), col='light blue')+ 
  geom_point()+ 
  stat_smooth(method="lm",col="red")

omartinez182

df <- data
Y <- 2 # 2 is the position of our "sales" column in the data frame
k <- 4 # k is equal to the size of the subset

#Array with all of the raw values of 'Y'
arr <- df[, Y]
#Empty array to store the values of the moving averages
MM<-rep(NA, length(arr))
index <- k-1
  

omartinez182

dfOut <- data.frame(df, MM=MM)

dfOut %>%
  ggplot(aes(x=t, y=sales))+
  geom_line(col = 'light blue')+
  geom_point()+
  geom_line(aes(x=t, y=MM), col = 'orange')+
  geom_point(aes(x=t, y=MM), col = 'orange')+
  xlab('Time')+
  ylab('Sales')+

omartinez182

df <- data
Y <- 2 #2 is the position of our "sales" column in the data frame
k <- 4 #k is equal to the size of the subset

#Array with all of the raws of 'Y'
arr <- df[, Y]
#Empty array to store the values of the moving averages
MM<-rep(NA, length(arr))
SI <- rep(NA, length(arr))
index <- k-1

omartinez182

dfOut %>%
  ggplot(aes(x=t, y=S))+
  geom_line(col="light blue")+
  geom_point(col="black")+
  xlab('Time')+
  ylab('Seasonality')+
  theme_minimal()

omartinez182

#Deseasonality
dfOut <- data.frame(dfOut, DES=(dfOut[ ,2]/dfOut[, 5])) #Y/S
dfOut %>%
  ggplot(aes(x=t, y=DES))+
  geom_line(col="light blue")+
  geom_point(col="black")+
  xlab('Time')+
  ylab('DES')+
  theme_minimal()

omartinez182

#Linear model
model <- lm(formula = DES ~ t, dfOut)

#Predictions
f <- predict(object = model, newdata = data.frame(t=dfOut[, 1]))
  
dfOut <- data.frame(dfOut, f=f)  

#New prediction accounting for seasonality
predictions <- dfOut[, 7]*dfOut[, 5]

omartinez182

MultForecast<-function(k, df, Y){
  #Array with all of the raw values of 'Y'
  arr <- df[, Y]
  #Empty array to store the values of the moving averages
  MM <- rep(NA, length(arr))
  SI <- rep(NA, length(arr))
  index <- k-1
  
  #Calculate moving averages + SI
  for(i in c(1:length(arr))){