food_data.R

library(gtrendsR)
library(CausalImpact)
library(tidyverse)

plot_for_name <- function(food, time="today+5-y") {
  df <- gtrendsR::gtrends(food, geo="US", time=time)
  idx <- length(df$interest_over_time$date) - 6
  ggplot2::ggplot(df$interest_over_time, aes(x=date, y=hits)) + 
    ggtitle(paste0("Search volume for '", food, "'"), subtitle = "Source: Google Trends, US Data") + geom_line() +
    geom_vline(xintercept = df$interest_over_time$date[idx], color='red') +
    ggplot2::labs(x='Date', y='Relative Search Volume')
}

ci_for_name <- function(food, time="today+5-y") {
  df <- gtrendsR::gtrends(food, geo="US", time=time)
  n_points = length(df$interest_over_time$date) 
  idx <- n_points - 6
  food_ci = CausalImpact(df$interest_over_time$hits, 
             pre.period = c(35,idx - 1), 
             post.period = c(idx, n_points - 1)), 
             model.args = list(nseasons = 52))
  return(food_ci)
}