---
title: "GP practices"
output: 
  flexdashboard::flex_dashboard:
    orientation: rows
    theme: 
        version: 5
        bootswatch: minty
date: "2026-07-01"
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = F)
library(flexdashboard)
library(dplyr)
library(readr)
library(ggplot2)
library(crosstalk)
library(plotly)
library(PostcodesioR)
library(leaflet)
library(viridis)

options(scipen=8) # to stop scientific notation kicking in

thematic::thematic_rmd()

boards <- read_rds("data/boards.rds")

gps <- read_csv("data/practice_contact_details_20260401_opendata.csv") |>
        select(PracticeCode, 
               GPPracticeName, 
               PracticeListSize, 
               Postcode, 
               PracticeType, 
               HB) |>
         left_join(boards) |>
         select(-HB)


## create a crosstalk data object with the practice code as the key
shared_gps <- SharedData$new(gps, 
                            key = ~PracticeCode, 
                            group = "gp_practices_subset")

## make a filter
board_filter = filter_select(id = "board", label = "Pick a Health Board", ~HBName, sharedData = shared_gps)
```

Row {data-height=200}
-----------------

### Summary text

+ Scotland has `r nrow(gps)` practices
+ The largest (the `r gps[which(gps$PracticeListSize == max(gps$PracticeListSize)),]$GPPracticeName`) has `r max(gps$PracticeListSize)` patients
+ The smallest (the `r gps[which(gps$PracticeListSize == min(gps$PracticeListSize)),]$GPPracticeName`) has `r min(gps$PracticeListSize)` patients
+ The average practice size is `r round(mean(gps$PracticeListSize))`

### Filter
```{r}
board_filter
```

Row {data-height=800}
-----------------

### Summary plot

```{r}
my_plot <- shared_gps |>
  ggplot() +
  geom_density(aes(x = PracticeListSize, fill = HBName), alpha = 2/5) +
  scale_x_log10() +
  xlab("Total list size (log10)") +
  ylab("Total number of practices") +
  labs(fill="Health board name") 

ggplotly(my_plot)
```


### Fancy map

```{r message=FALSE, warning=FALSE}
gps_sm <- gps |>
  slice_sample(n = 100) |>
  rowwise() |>
  mutate(postcode_look = postcode_lookup(Postcode)) |>
  tidyr::unnest(postcode_look) |>
  select(PracticeCode, GPPracticeName, PracticeListSize, longitude, latitude, HBName) # making a small subset to keep things quick

shared_gps_sm <- SharedData$new(gps_sm, 
                            key = ~PracticeCode, 
                            group = "gp_practices_subset") # new, linked, shareddata

pal <- colorFactor(palette = viridis(length(unique(gps_sm$HBName))), domain = gps_sm$HBName)

shared_gps_sm |>
  leaflet() |>
  addTiles() |>
  addCircleMarkers(radius = ~PracticeListSize / 1000, 
                   color = ~pal(HBName))
```