-
Notifications
You must be signed in to change notification settings - Fork 0
/
app.R
174 lines (139 loc) · 4.85 KB
/
app.R
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
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
# Load packages
library(shiny)
library(readxl)
library(ggplot2)
library(dplyr)
library(ggfortify)
library(stats)
library(forecast)
# ui
ui <- fluidPage(
titlePanel("Sales Forecast App"),
sidebarLayout(
sidebarPanel(
h5("Please ensure your uploaded CSV data (< 5MB) contains
`Year`, `Month` and `Sales Amt` columns."),
h5("You can also use the app without uploading any data.
Just click the `Start forecast` button!"),
fileInput(inputId = "input_file", label = "Upload CSV data here",
accept = c("text/csv",
"text/comma-separated-values,text/plain",
".csv")),
actionButton(inputId = "run_button", label = "Start forecast")
),
mainPanel(
tableOutput(outputId = "table"),
plotOutput(outputId = "plot"),
textOutput(outputId = "mape"),
verbatimTextOutput(outputId = "boxtest", placeholder = T),
textOutput(outputId = "interpret_boxtest")
)
)
)
# server
server <- function(input, output) {
# Get uploaded or default data
get_file_or_default <- reactive({
if (is.null(input$input_file)) {
DATASET_PATH <- "dataset/Data.xlsx"
readxl::read_xlsx(DATASET_PATH, sheet = "Data")
} else {
read.csv(input$input_file$datapath, check.names = F)
}
})
# Function to preprocess df
preprocess_df <- function(df){
# Replace NaN with zero
df$`Sales Amt`[is.na(df$`Sales Amt`)] = 0
# Aggregate sales amount by year and month
df <- df[c("Year", "Month", "Sales Amt")]
df <- df %>% group_by(Year, Month) %>%
summarise(Monthly_Sales=sum(`Sales Amt`))
df
}
# Function to get the years in df
get_years <- function(df){
# Get the years
years <- unique(df$Year)
# Split dataset into train and test
start_train <- years[1]
end_train <- tail(years, 2)[1]
start_test <- tail(years, 2)[2]
c(start_train, end_train, start_test)
}
# Function to compute Holt-Winters
compute_holt_winter <- function(df, start_train, end_train, start_test){
df_train <- subset(df, Year <= end_train)
df_test <- subset(df, Year == start_test)
# Perform Holt-Winters
dfts_train <- df_train$Monthly_Sales %>%
ts(start = c(start_train, 1), end = c(end_train, 12), frequency = 12)
HoltWinters(dfts_train, seasonal = "multiplicative")
}
# Function to plot result
plot_result <- function(df, hw_for, start_train, end_train, start_test){
# Plot final result
dfts_test <- df_test$Monthly_Sales %>%
ts(start = c(start_test, 1), end = c(start_test, 12), frequency = 12)
dfts <- df$Monthly_Sales %>%
ts(start = c(start_train, 1), end = c(start_test, 12), frequency = 12)
plot(hw_for, ylim = c(0,2.3e+7), ylab = "Monthly Sales")
lines(dfts, lty = 2, col = "red")
legend(x = "topright", legend=c("Forecast", "Fitted", "Original"),
col=c("blue", "black", "red"), lty=1:2, cex = 0.6)
}
get_head_df <- eventReactive(input$run_button, {
head(get_file_or_default())
})
get_df <- eventReactive(input$run_button, {
get_file_or_default()
})
eval_boxtest <- function(hw_for){
Box.test(hw_for$residuals, type="Ljung-Box")
}
eval_mape <- function(hw, df, start_test){
df_test <- subset(df, Year == start_test)
hw.pred <- predict(hw, n.ahead = 12,
prediction.interval = TRUE, level = 0.95)
actual <- df_test$Monthly_Sales
mean(abs((actual - hw.pred)/actual))*100
}
# Render df
output$table <- renderTable(get_head_df())
# Render plot
output$plot <- renderPlot({
df <- get_df()
df <- preprocess_df(df)
year_list <- get_years(df)
hw <- compute_holt_winter(df, year_list[1], year_list[2], year_list[3])
hw_for <- forecast(hw, h = 36, level = 0.95)
plot_result(df, hw_for, year_list[1], year_list[2], year_list[3])
})
# Render MAPE
output$mape <- renderText({
df <- get_df()
df <- preprocess_df(df)
year_list <- get_years(df)
hw <- compute_holt_winter(df, year_list[1], year_list[2], year_list[3])
mape <- eval_mape(hw, df, year_list[3])
paste("The mean absolute percentage error (MAPE) is", mape, ".")
})
# Render box test
output$boxtest <- renderPrint({
df <- get_df()
df <- preprocess_df(df)
year_list <- get_years(df)
hw <- compute_holt_winter(df, year_list[1], year_list[2], year_list[3])
hw_for <- forecast(hw, h = 36, level = 0.95)
eval_boxtest(hw_for)
})
# Render box test interpretation
output$interpret_boxtest <- renderText({
"The Ljung-Box test checks if the residual is independent
(residual is supposed to be independent).
Residual is independent or random if p-value > 0.05, i.e.
we fail to reject the null hypothesis."
})
}
# shiny object
shinyApp(ui = ui, server = server)