SDG_hackathon_notebook.Rmd

---
title: "SDG Hackathon"
author: "Kai Lim"
date: "05/11/2021"
output: html_document
editor_options: 
  chunk_output_type: inline
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

The goal of this data visualization hackathon is twofold:

1. To better understand the commonalities and differences between the five query-based SDG labeling systems.

2. To characterize the efforts of the Swiss research landscape to address the UN SDGs.

```{r load libraries, message=FALSE, warning=FALSE}
library(tidyverse)
library(lubridate)
library(ggalt) #needed for geom_dumbbell
library(patchwork)
library(magick)
library(cowplot)
library(ggthemes)
```

```{r load csvs}
sdg_data<-read.csv("sdg_hackathon_data.csv")
sdg_data_supp<-read.csv("supplementary_data.csv")
```

## Which type of SDG a project is in will be detected more easily by all sources?

```{r}
sdg_data_sources <- sdg_data %>%
  select(project_number,start_date,end_date,sdg,system, approved_amount) %>% 
  transform(start_date=as_date(start_date), end_date=as_date(end_date)) %>% 
  mutate(duration=as.period(interval(ymd(start_date),ymd(end_date)),unit="days")@day) %>% 
  filter(sdg!="SDG-17", approved_amount!="data not included in P3") %>%  #filter SDG-17 because some sources don't have it
  group_by(sdg,system) %>% 
  mutate(count=n()) %>% 
  mutate(med_duration=median(duration)) %>% 
  mutate(med_cost=median(as.numeric(approved_amount))) %>% 
  mutate(system=recode(system, 
         aurora="Aurora",
         elsevier="Elsevier",
         ontology="Ontology",
         sdsn="SDSN",
         siris="Siris"))

table(sdg_data$system)


SDG_names <- c(`SDG-01` = "1: No Poverty",
               `SDG-02` = "2: Zero Hunger",
               `SDG-03` = "3: Good Health & Well Being",
               `SDG-04` = "4: Quality Education",
               `SDG-05` = "5: Gender Equality",
               `SDG-06` = "6: Clean Water & Sanitation",
               `SDG-07` = "7: Affordable Clean Energy",
               `SDG-08` = "8: Decent Work & Economy Growth",
               `SDG-09` = "9: Industry, Innovation & Infrastructure",
               `SDG-10` = "10: Reduced Inequalities",
               `SDG-11` = "11: Sustainable Cities",
               `SDG-12` = "12: Responsible Consumption & Production",
               `SDG-13` = "13: Climate Action",
               `SDG-14` = "14: Life Below Water",
               `SDG-15` = "15: Life On Land",
               `SDG-16` = "16: Peace & Justice")

plot1<-sdg_data_sources %>% 
  ggplot(aes(x=system,y=count,color=system))+
  geom_point(aes(size=med_cost))+
  geom_segment(aes(x=system, xend=system, y=0, yend=count)) +
  facet_wrap(~sdg,labeller = as_labeller(SDG_names, label_wrap_gen(width=20)))+
  theme_bw() +
  labs(title="Number of projects identified by each system",
       color="Query system",
       size="Median approved funding\n(CHF)",
       caption="Github:kai-lim")+
  theme(legend.position="right",
        axis.text.x = element_blank(),
        axis.ticks.x = element_blank(),
        axis.line.x = element_blank())+
  scale_y_continuous(name="Number of projects", limits=c(0,20000))+
  scale_x_discrete(name="Query system used to detect SDGs")+
  theme(strip.background = element_rect(fill=c("#CCB7AE")))

plot1 

# tried to add image but didn't work - can try after deadline.
# ## read PNG file from web
# png <- magick::image_read("~/OneDrive - King's College London/PhD/outside_PhD/SDG_Hackathon/Hackathon_project_folder/SDG_title.png")
# ## turn image into `rasterGrob`
# img <- grid::rasterGrob(png, interpolate = TRUE)
# 
# (plot1_1 <- plot1 +
#   annotation_custom(img, ymin = 21.5, ymax = 30.5, xmin = 55, xmax = 65.5))


ggsave(file="SDG_2.png", width=8,height=8)

```

Looking good, can consider changing the y-axis to cost or med_ instead. 
change med_duration to cost perhaps?
Do it tomorrow. 
Consider using BBC's theme. 

add a second axis: https://www.r-graph-gallery.com/line-chart-dual-Y-axis-ggplot2.html 


```{r}
funding_data<-sdg_data_sources %>% 
  group_by(system, sdg) %>%
  mutate(approved_amount=as.numeric(approved_amount)) %>% 
  top_n(1, wt=approved_amount)%>% 
    mutate(approved_amount_K=as.numeric(approved_amount/1000))

# First ensure that the data feame is ordered by the start time
#funding_data <- funding_data[order(funding_data$start_date),]

# Now iterate through each row, calculating how many previous rows have
# earlier starts but haven't yet finished when the current row starts.
# Multiply this number by a small negative offset and add the 1.48 baseline value
#funding_data$offset <- 1.48 - 0.03 * sapply(seq(nrow(funding_data)), function(i) {
#   with(funding_data[seq(i),], length(which(start_date < start_date[i] & end_date > start_date[i])))
#   })


plot2<-funding_data %>% 
  ggplot(aes(x=start_date, xend=end_date,y=as.numeric(reorder(approved_amount_K,start_date))))+
  geom_segment(aes(color=system,yend=as.numeric(reorder(approved_amount_K,start_date))),
              size = 2,
              position = position_nudge(x = 0, y = 0),
              lineend = "round") +
  facet_wrap(~sdg)+
  scale_y_continuous(name="Approved amount (in thousands)")+
  theme_minimal()+
  labs(title="Duration of projects with highest funding detected by each query system",
       color="Query system used to detect SDGs",
       x="Project start and end date")+
  theme(legend.position = "bottom")

plot2
```



```{r both plots tgt}
plot1/plot2
```