#### RFM analysis #####
library(dplyr)
# filter to only United Kingdom
data <- data[data$Country == "United Kingdom", ]
data$PxQ <- data$Quantity * data$UnitPrice
data$Recency <- difftime(as.Date(max(data$InvoiceDate)), as.Date(data$InvoiceDate), units = "days")

RFMresult <- data %>%
  select(InvoiceNo, CustomerID, Recency, PxQ) %>%
  group_by(CustomerID) %>%
  summarise(Recency = min(Recency), Frequency = n_distinct(InvoiceNo), Monetary = sum(PxQ)/n_distinct(InvoiceNo))
RFMresult$Recency <- as.integer(RFMresult$Recency)