01_workshop_script.qmd

---
title: "ggplot2 and survey data" 
format: html 
editor: visual 
---

## 1. Load packages and import data

```{r}

# load packages
library(tidyverse) 
library(scales)  

# import data
results <- read_csv("employee_survey_results.csv")

```

## 2. Aggregate

```{r}

# counts and percentages
counts_perc <- results |> 
  count(statement, response, response_num) |> 
  group_by(statement) |> 
  mutate(total = sum(n), 
         percent = n / total) 

# medians
medians <- results |>    
  group_by(statement) |>    
  summarize(med = median(response_num)) # calculate median by statement
```

## 3. 100% stacked bar chart

#### 3a. Try a bare bones plot

```{r, fig.width= 8, fig.height=4}




```

#### 3b. Set sorting order

```{r}

# determine order of the bars (combined value of agree/strongly agree) 
stacked_order <- results |> 
  filter(response %in% c("Agree", "Strongly Agree")) |>
  count(statement) |> 
  arrange(n) |> 
  pull(statement) # converts column values to a vector

# set order of the bars and fill categories
counts_perc <- counts_perc |> 
  mutate(statement = factor(statement,
                            levels = stacked_order),
         response = factor(response,
                           levels = c("Strongly Agree",
                                      "Agree",
                                      "Neutral",
                                      "Disagree",
                                      "Strongly Disagree")))
```

#### 3c. Assign colors to response values

```{r}

fill_colors <- c(
  "Strongly Agree"    = "#3e6488",
  "Agree"             = "#829db2",
  "Neutral"           = "#c8cdd1",
  "Disagree"          = "#edad89",
  "Strongly Disagree" = "#e36c32"
)
```

#### 3d. Better plot

```{r, fig.width= 8, fig.height=4}

counts_perc |> 
  ggplot(aes(x = percent, y = statement, fill = response)) +
  geom_col(width = .7) +
  labs(x = NULL, y = NULL, fill = NULL) +
  scale_x_continuous(expand = c(0,0), 
                     labels = percent) +
  scale_fill_manual(values = fill_colors,
                    guide = guide_legend(reverse = TRUE)) +
  theme_minimal() +
  theme(legend.position = "top",
        panel.grid = element_blank(),
        panel.grid.major.x = element_line(color = "lightgrey"),
        plot.margin = margin(r = 50))
```

#### 3e. Save plot

```{r}
ggsave("stacked_bar.png", 
       width = 8, height = 4,
       dpi = 500, bg = "white")
```

## 4. Jittered strip plot

#### 4a. Try a bare bones plot

```{r, fig.height=4,fig.width=8}




```

#### 4b. Set sorting order

```{r}

# order medians data frame
medians <- medians |> 
  mutate(statement = fct_reorder(statement, med))

# order results data frame to match
results <- results |> 
  mutate(statement = factor(statement, 
                            levels = levels(medians$statement)))

```

#### 4c. Create a full grid to enclose jittered points

```{r}

# Get the unique values of x and y variables 
x_values <- 1:5 
y_values <- unique(results$statement)  

# Create a complete grid with all possible combinations 
full_grid <- expand.grid(   
  response_num = x_values,   
  statement = y_values 
  )
```

#### 4d. Better plot

```{r, fig.height=4,fig.width=8}

ggplot() +  
  geom_tile(full_grid,
            mapping = aes(x = response_num, 
                          y = statement),
            fill = NA,                                    
            color = "#dddddd",                              
            height = .8, 
            linewidth = .5) +
  geom_jitter(results, 
              mapping = aes(x = response_num, 
                            y = statement,
                            fill = "One respondent"),
              size = 2,
              height = .25,
              width = .25, 
              alpha = .5, 
              color = "#84B6E4") + 
  geom_point(medians,
             mapping = aes(x = med, 
                           y = statement,
                           color = "Median response"),
             shape = 108, 
             size = 11.5) +
  guides(color = guide_legend(override.aes = list(size = 6))) +
  labs(x = NULL, y = NULL, color = NULL, fill = NULL) +
  scale_color_manual(values = "#3e6488") +
  scale_x_continuous(expand = c(0,0), 
                     breaks = 1:5,                      
                     labels = c("1\nStrongly Disagree",
                                "2",
                                "3\nNeutral",
                                "4",
                                "5\nStrongly Agree")) + 
  theme_minimal() +   
  theme(panel.grid = element_blank(),
        axis.text.x = element_text(size = 10, lineheight = 1.1),
        legend.text = element_text(size = 10))
```

#### 4e. Save plot

```{r}
ggsave("jittered_strip.png", 
       width = 8, height = 4,
       dpi = 500, bg = "white")
```

## 5. Distribution heatmap

#### 5a. Try a bare bones plot

```{r, fig.height=4,fig.width=8}





```

#### 5b. Set sorting order

```{r}

# sort counts data frame based on median factor
counts_perc <- counts_perc |> 
  mutate(statement = factor(statement,
                            levels = levels(medians$statement)))
```

#### 5c. Better plot

```{r, fig.height=4,fig.width=8}

ggplot() +  
  geom_tile(counts_perc,
            mapping = aes(x = response_num, 
                          y = statement, 
                          fill = n),
            height = .8) +
  geom_tile(full_grid,
            mapping = aes(x = response_num,
                          y = statement),
            fill = NA,
            color = "#dddddd",
            height = .8,
            linewidth = .5) +
  geom_point(medians,
             mapping = aes(x = med, y = statement,
                           color = "Median response"),
             size = 4.3, shape=21, fill = NA) +
  labs(x = NULL, y = NULL, fill = "Count", color = NULL) +   
  scale_x_continuous(expand = c(0,0), breaks = 1:5,                      
                     labels = c("1\nStrongly Disagree",
                                "2",
                                "3\nNeutral",
                                "4",
                                "5\nStrongly Agree")) +  
  scale_color_manual(values = "black") +
  scale_fill_gradient(low  = "#EBF5FB",
                      high = "#3f739b") +
  theme_minimal() +   
  theme(panel.grid = element_blank(),
        axis.text.x = element_text(size = 10, lineheight = 1.1))
```

#### 5d. Save plot

```{r}
ggsave("dist_heatmap.png", 
       width = 8, height = 4,
       dpi = 500, bg = "white")
```