analysis.qmd

---
title: "analysis"
editor: visual
---

October 22, James Hoffman, youtuber and 201x World Barista Champion, conducted a 5,000 cupping session.

So, who are these coffee connoisseurs? Conducted 4 roasts: \* Natural Process

| A               | B   | C   | D   |
|-----------------|-----|-----|-----|
| pic             |     |     |     |
| Natural Process |     |     |     |
|                 |     |     |     |

```{r}
library(dplyr)
library(ggplot2)
```

```{r}
df = read.csv("data/GACTT_RESULTS_ANONYMIZED_v2.csv") |> janitor::clean_names()

df = df |> mutate_at(c("what_is_your_age", "gender", "education_level", "ethnicity_race", "political_affiliation"), ~ifelse(. == "", "Not Reported", .))
                
                #~tidyr::replace_na(., replace = "Unk"))
names(df)
```

```{r}
# TODO pop pyramid
df |> mutate(gender = ifelse(gender == "Prefer not to say", "Not Reported", gender)) |> count(what_is_your_age, gender) |> ggplot() + geom_col(aes(x = gender, y = n)) + facet_wrap(~what_is_your_age)

df |> mutate(gender = ifelse(gender == "Prefer not to say", "Not Reported", gender)) |> count(what_is_your_age, gender) |> mutate(
    population = ifelse(gender=="Male", n*(-1),
                        n*1)) %>%
    ggplot(aes(x = what_is_your_age,y = population, fill = gender)) + 
    geom_bar(stat = "identity") +
    coord_flip() #+ facet_wrap(~gender)
```

```{r}
df |> summary()
```