IPS pageviews

library(readr)
library(dplyr)
library(lubridate)
library(ggplot2)
library(tidyr)

data <- read_csv("~/traffic_2018-Jun-20_2018-Oct-18.csv")

data <- data %>%
  mutate(Week = lubridate::isoweek(date),
         DayOfWeek = weekdays(date))

data %>%
  filter(DayOfWeek == "Wednesday") %>%
  ggplot(aes(x = Week, y = `page views`)) +
  geom_line()

daysData <- data %>%
  select(Week, DayOfWeek, `page views`) %>%
  spread(DayOfWeek, `page views`) %>%
  select(Week, Monday, Wednesday) %>%
  filter(!is.na(Monday) & !is.na(Wednesday))

daysData %>% 
  mutate(RatioWedToMon = Wednesday / Monday) %>% 
  select(Week, Monday, Wednesday, RatioWedToMon)

#Exclude current week from data
train <- daysData %>% filter(Week != 42) 


train %>%
  transmute(Ratio = Wednesday / Monday) %>%
  summary


model <- lm(data = train, formula = Wednesday ~ Monday)

summary(model)

expected <- predict(model, daysData %>% filter(Week == 42))

actual <- daysData %>%
  filter(Week == 42) %>%
  select(Wednesday)