---
title: "Reports with tidyverse tools"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Reports with tidyverse tools}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(collapse = TRUE, comment = "#>")
library(dplyr)
source(system.file("recipes", "reports.R", package = "sensortowerR"))
```

The package fetches and normalizes data. Standard tidyverse tools handle analysis
and presentation. These executable examples use synthetic data, not API estimates.

```{r}
data <- tibble::tibble(
  app_id = "example", os = "unified", country = "US",
  date = as.Date(c("2025-01-01", "2026-01-01")),
  metric = "revenue", value = c(100, 150), unit = "USD", period = "month"
)
recipe_yoy(data)
recipe_portfolio(data)
```

YoY matches the same period one year earlier. Missing or zero baselines produce
missing growth. Portfolio totals reject duplicate observations and audience
metrics, whose users can overlap. Missing sales values keep totals missing.
Check coverage before interpreting complete totals; a successful response can
still omit small apps or unavailable periods.

```{r, eval=FALSE}
recipe_plot(data)
recipe_dashboard(data)
scales::label_dollar()(data$value)
```

For launch curves, add your chosen launch date and derive the day offset with
`mutate(day = as.integer(date - launch_date))`. Arrange within app and country
before `cumsum()`. Do not use `na.rm = TRUE` to hide missing launch observations.