merge.Rmd

---
title: "Integrating Conflict Data"
output:
  workflowr::wflow_html:
    toc: false
editor_options:
  chunk_output_type: console
---

This page shows how to merge Geo-PKO data with conflict data and visualise the results. The examples used here are Uppsala University's Violence Early Warning System (ViEWS) project, which forecasts conflict risk, and the Uppsala Conflict Data Programme (UCDP), one of the world's leading sources of data on armed conflict. Merging these datasets can provide insights into the links between conflict risk and peacekeeping deployments, and help policymakers make effective peacekeeping decisions where the risk of conflict is high.  

## Setting up

Load packages.
```{r, warning=FALSE, message=FALSE}
library(dplyr)
library(sp)
library(tidyr)
library(geojsonio)
library(broom)
library(rgdal)
library(ggplot2)
library(leaflet)
library(sf)
library(spdep)
library(maptools)
library(plyr)
library(rjson)
library(RJSONIO)
library(rmapshaper)
library(htmltools)
library(htmlwidgets)
library(leaflet.providers)
library(kableExtra)
```

## Geo-PKO x ViEWS

This visualisation aims to show peacekeeping deployments and projected conflict risk for the following 36 months. It helps us visualise conflict forecasts for locations where peacekeepers are deployed and better understand the relationship between peacekeeping deployments and conflict risk (for example, the risk levels at which peacekeeping deployments are more likely).

First, we import the datasets. We're using the published Geo-PKO dataset, and conflict forecast data from ViEWS for state-based conflict, non-state conflict, and one-sided violence over the next 36 months in Africa. Both datasets offer insights into the sub-national level, using a unique "PRIO-grid" identifier. The PRIO-grid is an innovative geospatial unit from the Peace Research Institute Oslo that divides the world into roughly 100km x 100km squares, allowing geographic analysis beyond the country level to be streamlined.  

```{r, warning=FALSE, message=FALSE}
library(readr)
geopko <- readr::read_csv("C:/Users/tanus/Dropbox/Geo-PKO working material for v 2.0/Dataset v.2.0/Geo_PKO_v.2.0.csv") 
#unzip("data/ViEWS.zip", exdir="data/ViEWS")
predictors <- read.csv("C:/Users/tanus/Documents/R/GeoPKO/data/ViEWS/ensemble_pgm.csv")
```

The Geo-PKO dataset includes detail on troop deployment numbers, types of troops, non-troop deployments, and contributing countries.  
 
```{r, warning=FALSE, message=FALSE}
kable(geopko[9546:9550,]) %>% kable_styling() %>%
  scroll_box(width = "100%", height = "200px")
```

### Preparing the data

The predictor database begins with July 2020 and forecasts the risk of conflict over the next 36 months ahead. Here's a preview of the data within it, showing state-based (`sb`), non-state (`ns`) and one-sided violence (`os`) forecasts. The variable `month_id` codes months differently to the Geo-PKO dataset, with every month assigned a different numeric value.  

```{r, warning=FALSE, message=FALSE}
kable(predictors[90545:90550,]) %>% kable_styling() %>%
  scroll_box(width = "100%", height = "200px")
```
  
You'll see both the Geo-PKO and ViEWS datasets include the PRIO-grid identification variable (`pg_id or prioid`), which corresponds to a specific grid square on the map. This is what we'll use to eventually merge the datasets.

We first need to filter the Geo-PKO data to include only deployments over the relevant time period. We are choosing to visualise deployments between July 2019 - June 2020 and conflict risk for the following 36 months (2020-2023). We also remove countries outside Africa, as the ViEWS dataset only covers this continent.

_Note: 2020 data is not currently available, but will be in early 2021. The dataset used for this visualisation will then be updated._

```{r, warning=FALSE, message=FALSE}
# filtering for troop deployments over July 2019 - June 2020 in Africa

geopko2 <- geopko %>%
  select(mission, month, year, prioid, no.troops, country, location, latitude, longitude) %>%
  dplyr::mutate_at(vars(longitude, latitude, month, year, no.troops), as.numeric) %>%
  filter(year==2019 & month>6 | year==2020 & month <7) %>%
  filter(country != "Kosovo" & country != "Lebanon" & country != "Haiti" & country != "Syria" & country != "Cyprus")

```

We now need to calculate the average number of troops deployed over that time period. In the visualisation, we'll use coordinates to map deployment locations, so the average is calculated by location. The following script also shows how to end up with an average number of deployed troops for each PRIO-grid square, which we won't visualise in this example, but is useful to understand how to merge the data.

```{r, warning=FALSE, message=FALSE}
# calculating an average number of troops per location

geopko3 <- geopko2 %>% 
  group_by(location) %>%
  filter (! duplicated(no.troops))  %>%
  dplyr::mutate(no.troops = mean(no.troops, na.rm=TRUE)) %>%
  dplyr::mutate(no.troops=round(no.troops)) %>%
  filter (! duplicated(no.troops))  %>%
  ungroup() %>%
  select(-month)

# summing troop count by priogrid id

geopko4 <- geopko3 %>%
  group_by(prioid) %>%
  dplyr::mutate(no.troops = sum(no.troops, na.rm=TRUE)) %>%
  filter (! duplicated(no.troops))  %>%
  ungroup() %>%
  select(-location)


```

Next, we can calculate the average conflict risk (state-based, non-state, and one-sided) for each PRIO-grid square over the time period we're provided with.

```{r, warning=FALSE, message=FALSE}

predictors2 <- predictors %>% 
  group_by(pg_id) %>%
  dplyr::mutate(average_allwthematic_sb = mean(average_allwthematic_sb, na.rm=TRUE)) %>%
  dplyr::mutate(average_allwthematic_ns = mean(average_allwthematic_ns, na.rm=TRUE)) %>%
  dplyr::mutate(average_allwthematic_os = mean(average_allwthematic_os, na.rm=TRUE)) %>%
  filter (! duplicated(pg_id)) %>%
  ungroup() %>%
  select(-month_id)

```

Finally, we merge the two datasets.

```{r, warning=FALSE, message=FALSE}
# merging geopko with conflict forecast data (averaged)

geopko4$pg_id <- geopko4$prioid

priogriddf <- full_join(
  geopko4, predictors2,
  by = c("pg_id"), 
  na.rm = TRUE)
  
```


### Merging the dataframe with the shapefile

Like we mentioned before, the PRIO-grid unit involves dividing the entire world into roughly 100km x 100km squares. That means that if we want to map it, we'll be working with large files, so keep that in mind when you're reading in the shapefile.  

```{r, message=FALSE, warning=FALSE}
shapefile <- rgdal::readOGR("C:/Users/tanus/Documents/R/R Files/ViEWS/priogrid.geojson")
```

The shapefile contains both geospatial polygon data and numerical data that corresponds to the ViEWS dataset; specifically, a PRIO-grid ID and a country ID. To work with the data within this shapefile, we need to fortify it. This converts it into a dataframe. We also convert the IDs to rownames to make it easier to work with. And, finally, we merge it with `pgnewdf2`, which we created earlier.  

```{r, warning=FALSE, message=FALSE}

# fortify
shapefile@data$id <- rownames(shapefile@data)
shapefile.df <- fortify(shapefile, region = "id")

# merge data
shapefile.df <- merge(shapefile, priogriddf, by.x = "pg_id", by.y = "pg_id", all.x=F, all.y=T, duplicateGeoms=TRUE)

```

Now `shapefile.df` has the new attributes, including variables from both Geo-PKO and ViEWS. `id` is a variable that ties the 'polygon', or a single square on the grid, to its location on the map. As many PRIO-grid squares will have a conflict forecast index but no troop deployments, many data points under Geo-PKO variables will show NA.
 
```{r, warning=FALSE, message=FALSE}
kable(shapefile.df@data[5386:5390,]) %>% kable_styling() %>%
  scroll_box(width = "100%", height = "200px")
```

### Mapping ViEWS data with Geo-PKO

To map the data, we're going to use the `leaflet` package (and a bunch of others to support it). The first thing we do is set up our colour palette and bins. We're using the 'viridis' colour palette, designed for accessibility and continuous-scale representation. The other thing we include is a small segment of code that fixes spacing between any NA value in the legend, and the remainder of the legend, making it easier to see.  

```{r, warning=FALSE, message=FALSE}
bins <- c(0, 10, 20, 50, 100, 200, 500, 1000, Inf)
pal <- colorNumeric("viridis", NULL)

#to fix spacing of NA in legend
css_fix <- "div.info.legend.leaflet-control br {clear: both;}" # CSS to correct spacing
html_fix <- htmltools::tags$style(type = "text/css", css_fix)  # Convert CSS to HTML
```

Next, let's map. We include three colour layers to shade squares according to their conflict forecast value. These layers can be toggled between state-based conflict, non-state conflict, and one-sided violence. Troop deployment numbers are included as labels, which you can see for each square on hover. Labels for specific deployment locations also show the name of the location and country, on hover over location markers. 

_Zoom in for more detail._

```{r, warning=FALSE, message=FALSE}
map <- leaflet(shapefile.df) %>%
  addTiles() %>%
  addPolygons(
    color = "#444444", weight = 0.25, smoothFactor = 0.5,
    opacity = 0.05, fillOpacity = 0.4,
    fillColor = ~ pal(shapefile.df@data$average_allwthematic_sb),
    group = "State-Based Conflict",
    highlightOptions = highlightOptions(
      color = "white", weight = 2,
      bringToFront = TRUE
    )
  )  %>%
  addPolygons(
    color = "#444444", weight = 0.25, smoothFactor = 0.5,
    opacity = 0.05, fillOpacity = 0.4,
    fillColor = ~ pal(shapefile.df@data$average_allwthematic_ns),
    group = "Non-State Conflict",
    highlightOptions = highlightOptions(
      color = "white", weight = 2,
      bringToFront = TRUE
    )
  ) %>%
  addPolygons(
    color = "#444444", weight = 0.25, smoothFactor = 0.5,
    opacity = 0.05, fillOpacity = 0.4,
    fillColor = ~ pal(shapefile.df@data$average_allwthematic_os),
    group = "One-Sided Violence",
    highlightOptions = highlightOptions(
      color = "white", weight = 2,
      bringToFront = TRUE
    )
  ) %>%
  # these circle markers use geopko3 data to cover multiple locations per prioid
  addCircleMarkers((data <- geopko3$no.troops),
    lat = ~latitude, lng = ~longitude,
    weight = 1, radius = 2, fillOpacity = 0.6, color = "darkblue",
    group = "Deployment Locations",
    label = paste(
      "<strong>Deployment Location</strong>", "- ", geopko3$mission,
      "<br/><strong>Number of Troops: </strong>", geopko3$no.troops,
      "<br/><strong>Location: </strong>", geopko3$location,", ", geopko3$country
    )
  %>%
  lapply(htmltools::HTML) 
  ) %>%
  addLegend("bottomright",
    pal = pal,
    values = shapefile.df$average_allwthematic_sb,
    title = "Conflict Forecast",
    opacity = 1
  ) %>%
  addLayersControl(
    baseGroups = c("State-Based Conflict", "Non-State Conflict", "One-Sided Violence"),
    overlayGroups = c("Deployment Locations"),
    options = layersControlOptions(collapsed = FALSE)
  ) %>% 
  htmlwidgets::prependContent(html_fix) # legend NA fix

map

```

And here we have it: an interactive map to view the latest peacekeeping data and projected conflict risk over the following 36 months. You can zoom in to areas of higher conflict risk to see more detail, including deployment locations. Note that some marked locations still say 0 for troop count; this indicates a mission site where no troops are deployed, but support personnel such as UNPOL or UNMO are present.  

We can also do the same but with a world terrain basemap to identify, at a basic level, potential geographic impacts such as the greater concentration of conflict risk in mountainous areas, and the general deployment of peacekeepers in non-mountainous areas.

```{r, warning=FALSE, message=FALSE}

# adding provider tiles - replaces addTiles()
map <- map %>% addProviderTiles("Esri.WorldTerrain")

map
```

Extensions of this visualisation can be even more useful, particularly with a time-slider that can help us identify how the risk of conflict changes given peacekeeping deployments (and vice versa). That would be a valuable be a project for the future.  


# Geo-PKO x UCDP

Another useful dataset is from the Uppsala Conflict Data Programme, which offers insights into deaths from armed conflict. Looking at this data in conjunction with peacekeeping data can be useful to draw conclusions into peacekeeping given the severity of armed conflict, or lack thereof. We start by importing the "UCDP Georeferenced Event Dataset (GED) Global version 20.1", of which a small excerpt of 2011 is shown in the table below.  

```{r, warning=FALSE, message=FALSE}
UCDP <- read_csv("data/UCDP/ged201.csv")
kable(UCDP[9546:9550,]) %>% kable_styling() %>%
  scroll_box(width = "100%", height = "200px")
```

## Preparing the data
For the Geo-PKO data we only use certain variables, and calculate the average troop number by year. The UCDP data has different variables you could use. In this example we took `best`, which indicates "the best (most likely) estimate of total fatalities resulting from integer an event" and `deaths_civilians`, which tells us how many of this `best` variable were civilian deaths.  

```{r, warning=FALSE, message=FALSE}
# Preparing the Geo-PKO dataset
GeoPKO_dataUCDP <- geopko %>%
  select(mission, year, location, latitude, longitude, no.troops) %>% # Select only the variables you need
  mutate_at(vars(latitude, longitude, no.troops), as.numeric) %>%
  group_by(mission, year, location) %>%
  mutate(ave.no.troops = as.integer(mean(no.troops, na.rm = TRUE))) %>% # Sum the troop numbers by year through using a combination of group_by & mutate
  select(-no.troops) %>% # Deselect the previous troop number variable
  distinct() %>% # Delete any duplicate rows
  drop_na(ave.no.troops) %>% # Remove NAs from the average troop count
  filter(ave.no.troops > 0) # Exclude any troop numbers under the value of 0

# Preparing the UCDP dataset
UCDP_dataframe <- UCDP %>%
  select(year, longitude, latitude, where_coordinates, best, deaths_civilians) %>%
  drop_na(latitude, longitude) %>%
  group_by(year, where_coordinates) %>%
  mutate(best = as.integer(mean(best, na.rm = TRUE))) %>% # Take the mean of the "best" variable
  mutate(deaths_civilians = as.integer(mean(deaths_civilians, na.rm = TRUE))) %>% # Take the mean of the "deaths_civilians" variable
  filter(best > 0 & year >= 1995) %>% # Filter that data so that it only keeps rows for the years after 1995.
  distinct()
```

## Setting the aesthetics 
You can either set all the colours by hand, as seen in `pal3`, or use the "viridis" package to create a colour scale for you, as shown in `pal2`. For the UCDP data we used shades of red. The Geo-PKO data is mapped with viridis, which includes blue, green, and yellow.  

```{r, warning=FALSE, message=FALSE}
pal2 <- colorBin((viridis::viridis(10)), GeoPKO_dataUCDP$ave.no.troops, bins = c(1, 50, 100, 500, 1000, 2000, 4000, 8000))
pal3 <- colorBin(c("#700524", "#8d072e", "#981f42", "#ed4d3a", "#af516c", "#d19bab", "#dcb4c0"),
  UCDP_dataframe$best,
  bins = c(1, 50, 100, 500, 1000, 2000, 4000, 10000, Inf)
)
```

## Mapping UCDP data with Geo-PKO

Just as we used leaflet for the VIEWS map, we do the same here. Note that in this case we did not merge the two datasets into one dataframe. When hovering over the circles, more information will be provided on either the UN peacekeeping deployment or on the UCDP conflict-related deaths. It is possible to use the PRIO-grid ID instead, and both datasets include this variable.  

```{r, warning=FALSE, message=FALSE}
UCDP_Overview_Map <- leaflet() %>%
  addTiles(options = providerTileOptions(noWrap = TRUE)) %>%
  addMeasure(position = "bottomleft", primaryLengthUnit = "kilometers") %>% # Adds a widget that can measure distances between two places to the map
  clearMarkers() %>%
  clearShapes() %>%
  addLegend("topright",
    pal = pal3,
    values = UCDP$best,
    title = "Fatalities",
    opacity = 1
  ) %>%
  addLegend("topright",
    pal = pal2,
    values = GeoPKO_dataUCDP$ave.no.troops,
    title = "Peacekeepers",
    opacity = 1
  ) %>%
  addLayersControl(
    baseGroups = c("2019", "2018", "2015", "2010", "2005", "2000", "1995"),
    options = layersControlOptions(collapsed = FALSE), position = "topleft"
  ) %>%
  addCircleMarkers(
    data = (GeoPKO2019 <- GeoPKO_dataUCDP %>% filter(year == 2019)),
    color = ~ pal2(ave.no.troops), radius = ~ (ave.no.troops)^(1 / 3),
    opacity = 0.15, fillOpacity = 0.5,
    lng = ~longitude, lat = ~latitude, group = "2019",
    label = paste(
      "<strong>UN Peacekeeping Site<br/>Mission:</strong>", GeoPKO2019$mission,
      "<br/><strong>Location:</strong>", GeoPKO2019$location,
      "<br/><strong>Troops Deployed:</strong>", GeoPKO2019$ave.no.troops
    ) %>%
      lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (UCDP2019 <- UCDP_dataframe %>% filter(year == 2019)),
    lng = ~longitude, lat = ~latitude,
    color = ~ pal3(best), radius = ~ (best)^(1 / 3),
    opacity = 0.05, fillOpacity = 0.4,
    group = "2019",
    label = paste(
      "<strong>UCDP Reported Fatalities<br/>Location:</strong>", UCDP2019$where_coordinates,
      "<br/><strong>Total deaths:</strong>", UCDP2019$best,
      "<br/><strong>Civilian deaths:</strong>", UCDP2019$deaths_civilians
    ) %>%
      lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (GeoPKO2018 <- GeoPKO_dataUCDP %>% filter(year == 2018)),
    color = ~ pal2(ave.no.troops), radius = ~ (ave.no.troops)^(1 / 3),
    opacity = 0.15, fillOpacity = 0.5,
    lng = ~longitude, lat = ~latitude, group = "2018",
    label = paste(
      "<strong>UN Peacekeeping Site<br/>Mission:</strong>", GeoPKO2018$mission,
      "<br/><strong>Location:</strong>", GeoPKO2018$location,
      "<br/><strong>Troops Deployed:</strong>", GeoPKO2018$ave.no.troops
    ) %>%
      lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (UCDP2018 <- UCDP_dataframe %>% filter(year == 2018)),
    lng = ~longitude, lat = ~latitude,
    color = ~ pal3(best), radius = ~ (best)^(1 / 3),
    opacity = 0.05, fillOpacity = 0.4,
    group = "2018",
    label = paste(
      "<strong>UCDP Reported Fatalities<br/>Location:</strong>", UCDP2018$where_coordinates,
      "<br/><strong>Total deaths:</strong>", UCDP2018$best,
      "<br/><strong>Civilian deaths:</strong>", UCDP2018$deaths_civilians
    ) %>% lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (GeoPKO2015 <- GeoPKO_dataUCDP %>% filter(year == 2015)),
    color = ~ pal2(ave.no.troops), radius = ~ (ave.no.troops)^(1 / 3),
    opacity = 0.15, fillOpacity = 0.5,
    lng = ~longitude, lat = ~latitude, group = "2015",
    label = paste(
      "<strong>UN Peacekeeping Site<br/>Mission:</strong>", GeoPKO2015$mission,
      "<br/><strong>Location:</strong>", GeoPKO2015$location,
      "<br/><strong>Troops Deployed:</strong>", GeoPKO2015$ave.no.troops
    ) %>%
      lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (UCDP2015 <- UCDP_dataframe %>% filter(year == 2015)),
    lng = ~longitude, lat = ~latitude,
    color = ~ pal3(best), radius = ~ (best)^(1 / 3),
    opacity = 0.05, fillOpacity = 0.4,
    group = "2015",
    label = paste(
      "<strong>UCDP Reported Fatalities<br/>Location:</strong>", UCDP2015$where_coordinates,
      "<br/><strong>Total deaths:</strong>", UCDP2015$best,
      "<br/><strong>Civilian deaths:</strong>", UCDP2015$deaths_civilians
    ) %>% lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (GeoPKO2010 <- GeoPKO_dataUCDP %>% filter(year == 2010)),
    color = ~ pal2(ave.no.troops), radius = ~ (ave.no.troops)^(1 / 3),
    opacity = 0.15, fillOpacity = 0.5,
    lng = ~longitude, lat = ~latitude, group = "2010",
    label = paste(
      "<strong>UN Peacekeeping Site<br/>Mission:</strong>", GeoPKO2010$mission,
      "<br/><strong>Location:</strong>", GeoPKO2010$location,
      "<br/><strong>Troops Deployed:</strong>", GeoPKO2010$ave.no.troops
    ) %>%
      lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (UCDP2010 <- UCDP_dataframe %>% filter(year == 2010)),
    lng = ~longitude, lat = ~latitude,
    color = ~ pal3(best), radius = ~ (best)^(1 / 3),
    opacity = 0.05, fillOpacity = 0.4,
    group = "2010",
    label = paste(
      "<strong>UCDP Reported Fatalities<br/>Location:</strong>", UCDP2010$where_coordinates,
      "<br/><strong>Total deaths:</strong>", UCDP2010$best,
      "<br/><strong>Civilian deaths:</strong>", UCDP2010$deaths_civilians
    ) %>% lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (GeoPKO2005 <- GeoPKO_dataUCDP %>% filter(year == 2005)),
    color = ~ pal2(ave.no.troops), radius = ~ (ave.no.troops)^(1 / 3),
    opacity = 0.15, fillOpacity = 0.5,
    lng = ~longitude, lat = ~latitude, group = "2005",
    label = paste(
      "<strong>UN Peacekeeping Site<br/>Mission:</strong>", GeoPKO2005$mission,
      "<br/><strong>Location:</strong>", GeoPKO2005$location,
      "<br/><strong>Troops Deployed:</strong>", GeoPKO2005$ave.no.troops
    ) %>%
      lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (UCDP2005 <- UCDP_dataframe %>% filter(year == 2005)),
    lng = ~longitude, lat = ~latitude,
    color = ~ pal3(best), radius = ~ (best)^(1 / 3),
    opacity = 0.05, fillOpacity = 0.4,
    group = "2005",
    label = paste(
      "<strong>UCDP Reported Fatalities<br/>Location:</strong>", UCDP2005$where_coordinates,
      "<br/><strong>Total deaths:</strong>", UCDP2005$best,
      "<br/><strong>Civilian deaths:</strong>", UCDP2005$deaths_civilians
    ) %>% lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (GeoPKO2000 <- GeoPKO_dataUCDP %>% filter(year == 2000)), color = ~ pal2(ave.no.troops), radius = ~ (ave.no.troops)^(1 / 3),
    opacity = 0.15, fillOpacity = 0.5,
    lng = ~longitude, lat = ~latitude, group = "2000",
    label = paste(
      "<strong>UN Peacekeeping Site<br/>Mission:</strong>", GeoPKO2000$mission,
      "<br/><strong>Location:</strong>", GeoPKO2000$location,
      "<br/><strong>Troops Deployed:</strong>", GeoPKO2000$ave.no.troops
    ) %>%
      lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (UCDP2000 <- UCDP_dataframe %>% filter(year == 2000)),
    lng = ~longitude, lat = ~latitude,
    color = ~ pal3(best), radius = ~ (best)^(1 / 3),
    opacity = 0.05, fillOpacity = 0.4,
    group = "2000",
    label = paste(
      "<strong>UCDP Reported Fatalities<br/>Location:</strong>", UCDP2000$where_coordinates,
      "<br/><strong>Total deaths:</strong>", UCDP2000$best,
      "<br/><strong>Civilian deaths:</strong>", UCDP2000$deaths_civilians
    ) %>% lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (GeoPKO1995 <- GeoPKO_dataUCDP %>% filter(year == 1995)), color = ~ pal2(ave.no.troops), radius = ~ (ave.no.troops)^(1 / 3),
    lng = ~longitude, lat = ~latitude,
    opacity = 0.15, fillOpacity = 0.5,
    group = "1995",
    label = paste(
      "<strong>UN Peacekeeping Site<br/>Mission:</strong>", GeoPKO1995$mission,
      "<br/><strong>Location:</strong>", GeoPKO1995$location,
      "<br/><strong>Troops Deployed:</strong>", GeoPKO1995$ave.no.troops
    ) %>%
      lapply(htmltools::HTML)
  ) %>%
  addCircleMarkers(
    data = (UCDP1995 <- UCDP_dataframe %>% filter(year == 1995)),
    lng = ~longitude, lat = ~latitude, color = ~ pal3(best), radius = ~ (best)^(1 / 3),
    opacity = 0.05, fillOpacity = 0.4,
    group = "1995",
    label = paste(
      "<strong>UCDP Reported Fatalities<br/>Location:</strong>", UCDP1995$where_coordinates,
      "<br/><strong>Total deaths:</strong>", UCDP1995$best,
      "<br/><strong>Civilian deaths:</strong>", UCDP1995$deaths_civilians
    ) %>% lapply(htmltools::HTML)
  )

# to save as HTML, you can use the following code:
# saveWidget(UCDP_Overview_Map, file="geopko&ucdp - geopko.html")

# Shows the map
UCDP_Overview_Map
```

And here we have it again: an interactive map to view peacekeeping deployments, in a few selected years, and conflict-related deaths within that same year. We opted to allow the user to select different years to see how both the numbers and locations of both datasets change over time.


## Merging the UCDP and Geo-PKO data

If you do want to combine the two datasets into one dataframe, there are different ways to do so. As shown with the VIEWS data, you can merge the datasets based on certain variables using "full_join", causing the two datasets to be combined in one file. This creates a dataframe in which the variables of one of the dataset occur next to the other. This can be used for the UCDP data as well. However, you can also merge datasets "under" each other. For the UCDP we wanted to show this second way of merging. Variables that are the same between the two datasets might go by different names, which is why some of the UCDP variables are renamed so that they are in line with the Geo-PKO variables. The newly combined dataset will first show all the UCDP rows, and under these the Geo-PKO rows will be "pasted". 
Merging can be useful for all kinds of reasons, such as statistical analysis.  

```{r}
# Rename variables you want to merge
# UCDP_dataframeMerge = UCDP_dataframe %>% rename (location=where_coordinates)

# Merge the datasets
# Combined_UCDP_GeoPKO<- bind_rows(UCDP_dataframeMerge, GeoPKO_dataUCDP)
```