attempts at adding predictors.Rmd

---
title: "Adding other predictors"
author: "Andrew Baxter"
date: "14/10/2019"
output: html_document
---


```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE, 
                      message = FALSE, 
                      warning = FALSE,
                      error = FALSE,
                      fig.width = 6,
                      fig.height = 5,
                      dpi = 150,
                      out.width = "100%")
library(readxl)
library(tidyverse)
library(stringr)
library(broom)
library(magrittr)
library(Synth)
library(Rcpp)
library(purrr)
library(svglite)
library(SPHSUgraphs)
library(plotly)
library(knitr)
`-.gg` <- function(e1, e2) e2(e1)
load('Data/synth_data.rdata') # outputted from 'Synth_data.R'
load('Data/synth_data_b.rdata') # outputted from 'Synth_data.R'
load("Data/filtered_itsp.rdata")  # outputted from 'Synth_create_sps.R'
load("Data/placebo_country_b.rdata")  # outputted from 'Synth_create_sps.R'
load("Data/time_placebos_b.rdata")  # outputted from 'Synth_time_pbs.R'
source('R/Synth_functions.R')
```
```{r results='asis'}
cat("
<style>
caption {
      color: black;
      font-weight: bold;
      font-size: 1.0em;
    }
</style>
")

```


# Adding other predictors

```{r it_u18_gdp}
start_yr <- 1990

sd_new <- synthData_u20_filt[,c(1:4, 6)] %>% filter(Country!="Scotland") %>% 
  mutate(GDPperCap = GDPperCap/1000)

cc_new <- u_20_ccodes_f %>% filter(Country!="Scotland")

it_u20_gdp <- testSynthIterations(
  yrs = start_yr:1998,
  pred = "GDPperCap",
  data = sd_new,
  ccodes = cc_new,
  # data = synthData_u20[,c(1:4, 6)] %>% filter(!Country %in% exclude_u20_gdp),
  # ccodes = u_20_ccodes %>% filter(!Country %in% exclude_u20_gdp),
  n = 4,
  predictors = NULL,
  time.optimise = start_yr:1998,
  dependent = "pRate"
) %>%
  arrange(groups, mspe)

it_u20_pr <-  testSynthIterations(
  yrs = start_yr:1998,
  pred = "pRate",
  data = sd_new,
  ccodes = cc_new,
  n = 4,
  time.optimise = start_yr:1998
) %>% arrange(groups, mspe)

# dataprep(
#   foo = synthData_u20_filt[,c(1:4, 6)] %>% filter(Country!="Scotland"),
#   special.predictors = list(list("GDPperCap", 1990:1998, "mean")),
#   time.predictors.prior = 1990:1998,
#   dependent = "pRate",
#   time.plot = 1990:1998,
#   time.optimize.ssr = 1990:1998,
#   controls.identifier = u_20_ccodes_f %>% filter(Country != "England and Wales" & Country!="Scotland") %>% pull(Code),
#   treatment.identifier = u_20_ccodes_f %>% filter(Country == "England and Wales") %>% pull(Code),
#       unit.variable = "Code",
#       unit.names.variable = "Country",
#       time.variable = "Year"
# )

```

```{r}
it_u20_gdp %>% 
    group_by(groups) %>% 
  top_n(3, -mspe)  


it_u20_pr %>% 
    group_by(groups) %>% 
  top_n(3, -mspe)  
```

```{r}
sp_u20_gdp <- it_u20_gdp$sPred[it_u20_gdp$iteration == 4][[1]]
sp_u20_pr <- it_u20_pr$sPred[it_u20_pr$iteration == 4][[1]]


synthPrep(
  data = sd_new,
  "u20_gdp",
  dependent = "pRate",
  special.predictors = append(sp_u20_gdp, sp_u20_pr),
  time.optimise.ssr = start_yr:1998,
  time.predictors.prior = start_yr:1998,
  time.plot = start_yr:2013
)

```

```{r}
st_u20_gdp$tab.w %>% arrange(desc(w.weights)) %>% select(Country = unit.names, Weight = w.weights) %>% 
  kable(caption = "Country weights")

st_u20_gdp$tab.v %>% kable(caption = "Predictor weights")

st_u20_gdp$tab.pred %>% kable(caption = "Predictor balance between synthetic and treated units")
```

```{r}
gg_synth(md = md_u20_gdp, post = FALSE)
# gg_synth(md = md_u20_sp, post = FALSE)
```