---
title: "NYC Restaurant Inspections Dashboard: Analyses on a Subset of Asian Restaurants (Thai, Chinese, Indian, Korean, & Japanese)"
output: 
  flexdashboard::flex_dashboard:
    orientation: columns
    vertical_layout: fill
    source_code: embed
editor_options: 
  chunk_output_type: console
---

```{r setup, include=FALSE}
# Load packages for data and analysis
library(flexdashboard)
library(tidyverse)
library(p8105.datasets)
library(plotly)
library(lubridate)

# Load Instacart data
data("rest_inspec")

# Clean and sample restaurant inspections data
restaurant_sample = rest_inspec %>% 
  # Subset the data to a smaller sample by filtering for Asian cuisines
  filter(cuisine_description %in% c("Thai", "Chinese", "Indian", "Korean", "Japanese")) %>% 
  # Eliminate unnecessary columns %>% 
  select(dba, cuisine_description, boro, critical_flag, inspection_date, score, grade, street) %>% 
  # Rename columns
  rename(
    cuisine = cuisine_description,
    name = dba
  ) %>% 
  # Scores should not be less than 0
  filter(score > 0) %>% 
  # Drop NAs
  drop_na() %>% 
  # Parse inspection dates
  mutate(
    inspection_date = as.POSIXct(inspection_date, format = "%Y-%m-%d")
  ) %>% 
  mutate(
    inspection_month = month(inspection_date),
    inspection_year = year(inspection_date),
    inspection_month = month.abb[inspection_month]
  ) %>% 
  select(-inspection_date)

```

Column {data-width=650}
-----------------------------------------------------------------------

### Average inspection score over time by borough

```{r first plot (line plot, average score over time by borough)}
# Build a lineplot showing how the average inspection score has changed over time split by borough
restaurant_sample %>% 
  mutate(
    inspection_year = as.factor(inspection_year)
  ) %>% 
  # Group by borough and year
  group_by(boro, inspection_year) %>% 
  # Find mean per borough/year
  summarize(
    mean_score = mean(score)
  ) %>% 
  # Filter out when borough is missing
  filter(!boro == "Missing") %>% 
  # Generate lineplot
  plot_ly(x = ~inspection_year, y = ~mean_score,
          color = ~boro, type = "scatter", mode = "lines+markers", colors = "viridis") %>% 
  layout(
    title = "Mean Inspection Score Over Time By Borough",
    xaxis = list(title = "Inspection Year"),
    yaxis = list(title = "Average Score")
  )
```

Column {data-width=350}
-----------------------------------------------------------------------

### Number of restaurants per inspection grade in each borough

```{r second plot (bar plot, grade frequency per borough)}
# Build a bar plot for frequency of each grade per borough
restaurant_sample %>% 
  # Filter only for A's, B's, and C's
  filter(!(grade %in% c("Not Yet Graded", "P", "Z"))) %>% 
  # Find how many inspections (denominators)
  group_by(boro, grade) %>% 
  filter(!boro == "Missing") %>% 
  summarize(
    num_per_grade_boro = n()
  ) %>% 
  group_by(boro) %>% 
  # Find grade frequencies (numerators)
  mutate(
    total_grades = sum(num_per_grade_boro),
    grade_freq = num_per_grade_boro * 100 / total_grades
  ) %>% 
  # Plot stacked bar graph
  plot_ly(x = ~boro, y = ~grade_freq, 
          color = ~grade, type = "bar", colors = "viridis") %>% 
  layout(barmode = "stack") %>% 
  layout(
    title = "Frequency of Inspection Grade By Borough",
    xaxis = list(title = "Borough"),
    yaxis = list(title = "% Per Score Category")
  )
```

### Score distribution by cuisine

```{r third plot (boxplot, score by cuisine)}
# Generate boxplot for score distribution by cuisine
restaurant_sample %>% 
  # Reorder based on median score per cuisine
  mutate(
    cuisine = fct_reorder(cuisine, score)
  ) %>% 
  # Generate horizontal boxplot
  plot_ly(y = ~cuisine, x = ~score, color = ~cuisine,
          type = "box", colors = "viridis") %>% 
  layout(
    title = "Inspection Score Distribution Per Cuisine Across NYC Boroughs",
    xaxis = list(title = "Cuisine"),
    yaxis = list(title = "Inspection Score")
  ) %>% 
  # Hide legend
  layout(showlegend = FALSE)
```