3_explore-counts-longseries.Rmd

---
title: "Počty úředníků: 2003-2018"
author: "Petr Bouchal"
date: "11/2/2019"
output: html_document
execute:
  freeze: auto
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = T, warning = F, message = F, rows.print = 20)
library(readr)
library(dplyr)
library(ggplot2)
library(tidyr)
library(tibble)
library(lubridate)
library(purrr)
knitr::opts_chunk$set(echo = TRUE)
options(scipen = 6)
```

```{r}
c13o <- read_csv(here::here("data-output/legacy/groups_ALL.csv"))
dt0 <- read_rds(here::here("data-interim/objemy_pocty_scraped_raw_2012_2018.rds"))
```

```{r, max.print = 15}
unique(c13o$variable) %>% enframe(name = NULL)
unique(c13o$grp) %>% enframe(name = NULL)
unique(dt0$type) %>% enframe(name = NULL)
```

# Explore & prep to merge {.tabset}

```{r, include=F}
c13o %>% 
  select(grp, sgrp) %>% 
  distinct()
```

## Prep data 2013+

```{r}
dt <- dt0 %>% 
  filter(kap_num == "C E L K E M" & indicator == "count") %>% 
  select(year, grp = type, schvaleny = rozp,
         skutecnost, upraveny, rozdil, index, plneni) %>% 
  mutate(grp = recode(grp, 
                      `jedn. OSS státní správy` = "neústřední st. správa",
                      `ST.SPRÁVA` = "St. sprava se SOBCPO"),
         plneni = 2-plneni/100, rozdil = -rozdil)
```

## Prep data 2003+

```{r}
dto <- c13o %>% 
  filter(promenna == "Zam") %>% 
  select(grp, sgrp, variable, value, udaj, promenna, UO, exekutiva, Year) %>% 
  set_names(tolower(names(.))) %>% 
  mutate_at(vars(udaj, variable, promenna), tolower) %>% 
  mutate(year = year(year),
         udaj = recode(udaj,  upr2skut = "plneni",
                       uprminusskut = "rozdil"),
         grp = recode(grp, UO = "ÚO", `OSS-RO` = "OSS sum",
                      PO = "PO sum",
                      `OSS-SS` = "neústřední st. správa",
                      OOSS = "ostatní OSS")) %>% 
  filter(grp %in% c("ÚO", "ST.SPRÁVA", "SOBCPO", "OSS sum", "ostatní OSS",
                    "PO sum", "St. sprava se SOBCPO", 
                    "neústřední st. správa"))
```

## Check groupings {.tabset}

### 2003+

```{r}
unique(dto$grp) %>% enframe(name = NULL)
```

### 2013+ 

```{r}
unique(dt$grp) %>% enframe(name = NULL)
```

## Check data by comparing grouping sizes {.tabset}

### 2003+

```{r}
dto %>% 
  filter(udaj == "schvaleny" & year == 2013) %>% 
  ggplot(aes(grp, value/1000)) +
  geom_col() + coord_flip() + ggtitle("2013 - data 2003-2012") + 
  scale_y_continuous(limits = c(0,250))
```

### 2013+

```{r}
dt %>% 
  filter(year == 2013) %>% 
  filter(!(grp %in% c("Příslušníci a vojáci"))) %>%
  ggplot(aes(grp, skutecnost/1e3)) +
  geom_col() + coord_flip() + ggtitle("2013 - data 2013-2018") +
  scale_y_continuous(limits = c(0,250))
```

## Merge data:

```{r}
srs <- bind_rows(dto %>% mutate(ds = "old") %>%
                   filter(year != 2013) %>% 
                   select(year, grp, value, udaj),
                 dt %>% pivot_longer(names_to = "udaj", values_to = "value",
                                     cols = c(schvaleny, skutecnost,
                                              upraveny, rozdil, 
                                              index, plneni)) %>% 
                   mutate(ds = "new", year = as.numeric(year)))
```

# First charts {.tabset}

NB:

- skok v roce 2012 je daný redefinicí MV, které od 2012 zahrnuje i velení policie a hasičů
- řada příslušníků a vojáků začíná až v 2013, protože z minulé analýzy jsme je tuším úplně vypustil nebo v interních datech MF nebyli

## Absolute - comparable 

```{r}
srs %>% 
  filter(udaj == "skutecnost") %>% 
  ggplot(aes(year, value)) +
  geom_line() +
  facet_wrap(~grp)
```

## Absolute - focus on changes

```{r}
srs %>% 
  filter(udaj == "skutecnost") %>% 
  ggplot(aes(year, value)) +
  geom_line() +
  facet_wrap(~grp, scales = "free_y")
```

# Comparisons {.tabset}

## Plan vs. reality

```{r}
srs %>% 
  filter(udaj == "plneni") %>% 
  ggplot(aes(year, 2-value)) +
  geom_line() +
  facet_wrap(~grp)
```

## Growth from 2003 base

```{r}
srs %>% 
  filter(udaj == "skutecnost") %>% 
  group_by(grp) %>% 
  arrange(year) %>% 
  mutate(index = value/first(value)) %>% 
  ggplot(aes(year, index)) +
  geom_line() +
  facet_wrap(~grp)
```