AMZ_Projects/template.R at master · dfadeeff/AMZ_Projects · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85

rm(list = ls())

library(ggplot2)
library(forecast)
library(tseries)

mydata = read.table(file = "../../../../adhoc_queries/new/data/weekly_stats.txt",sep='\t', header = F)


generate_Fourier_ARIMA_forecast <- function(container, column, marketplace, horizon) {

  #function produces prediction for marketplace using Fourier and ARIMA model

  colnames(container) <- c("total_customers","new_customers",'ops','units','cogs','market','date')
  container$date = as.Date(container$date, format = "%Y-%m-%d")
  container$returning_customers <- container$total_customers - container$new_customers

  #Subset to get marketplace
  container <- container[ (container$market == marketplace), ]
  container <- container[ (container$date < Sys.Date() - 3), ]

  ggplot(data = container,aes(x = container$date, y = container$total_customers)) + geom_line()
  container <- container[(container$date > '2014-09-27'),]

  #check sorting
  container <- container[order(container$date),]

  #Initialise dependent var
  z = which(colnames(container) == column)[1]
  y = container[,z]

  #Create clean data series
  count_ts = ts(data = container[,z])
  container$clean_total <- count_ts

  h = horizon
  training <- window(container$clean_total, end = length(container$clean_total) - h)
  test <- window(container$clean_total, start = length(container$clean_total) - h + 1)
  count_bats <- tbats(ts(log10(training),freq=365.25/7)) #Here, important!
  pred <- forecast(count_bats, h=h)


  return(pred)
}

generate_preprocessed_dataframe <- function(container, column, marketplace, horizon) {

  #function produces prediction for marketplace using Fourier and ARIMA model

  colnames(container) <- c("total_customers","new_customers",'ops','units','cogs','market','date')
  container$date = as.Date(container$date, format = "%Y-%m-%d")
  container$returning_customers <- container$total_customers - container$new_customers

  #Subset to get marketplace
  container <- container[ (container$market == marketplace), ]
  container <- container[ (container$date < Sys.Date() - 3), ]

  ggplot(data = container,aes(x = container$date, y = container$total_customers)) + geom_line()
  container <- container[(container$date > '2014-09-27'),]

  #check sorting
  container <- container[order(container$date),]

  #Initialise dependent var
  z = which(colnames(container) == column)[1]
  y = container[,z]

  #Create clean data series
  count_ts = ts(data = container[,z])
  container$clean_total <- count_ts

  return(container)
}


pred = generate_Fourier_ARIMA_forecast(mydata,"new_customers",4,50)
mydata = generate_preprocessed_dataframe(mydata,"new_customers",4,50)


#Plot the prediction
plot(pred)

#Plot everything
autoplot(10^(pred$mean)) + autolayer(ts(mydata$clean_total,freq=365.25/7)) + autolayer(10^(pred$fitted))