_ialeuk25_analysis.Rmd

---
title: "analysis of ialeUK conference proceedings"
author: "James Millington"
output:
  html_document:
    number_sections: yes
    toc: yes
    code_folding: hide
    df_print: paged
    keep_md: yes
  pdf_document:
    highlight: tango
    number_sections: yes
    toc: yes
  word_document:
    toc: yes
linkcolor: red
citecolor: cyan
urlcolor: blue
---

# Total Conference Contributions

```{r results='hide', warning=F, message=F}
#Load Data
#(After slightly cleaning column titles - in future include code to do that here)

rm(list=ls())
library(tidyverse)
library(ggplot2)
path <- "C:/Users/k1076631/Google Drive/Research/Papers/InProgress/ialeUK_25years/QuantAnalysis/Rproject"
setwd(path)
filename <- "abstract_review_export_2018-06-11.csv"
cpdata <- read_csv(filename)
```

This document contains analysis by year. Future analysis could examine contribution attributes by:

- author affiliation (e.g. do NGOs conduct studies at particular scales?)
- landscape type (e.g. what species do studies in Urban landscapes focus on?) 
- species (e.g. are birds studied more using empirical studies or GIS?) 

etc.


```{r}
#spec(cpdata)

yrdata <- cpdata %>%
  select_if(is.numeric) %>%
  group_by(`Conference Year`) %>%
  summarise_all(sum, na.rm=T) 

```





Quick observations:

- general increase through time to early 2000s then drop but steady through 2010s
```{r}
authorCounts <- yrdata %>%
  select(`Conference Year`,Academic, Government,NGO,Business,Private) %>%
  mutate(yrsum = rowSums(.[2:6])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum)  #calculate proportion

ggplot(authorCounts, aes(x=`Conference Year`, y=count)) + geom_bar(stat="identity")

```


# Analysis by Conference Year

Stacked bar plots of contributions (by types and year)

## Author Affiliation

Quick observations:

- Academic contributors generally dominate
- Government contributors have decreased through time
- NGO attendance has replaced declines in Government? (could check sum of Gov + NGO through time)

```{r}
authorCounts <- yrdata %>%
  select(`Conference Year`,Academic, Government,NGO,Business,Private) %>%
  mutate(yrsum = rowSums(.[2:6])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum)  #calculate proportion


ggplot(authorCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(authorCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(authorCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")


```

## Landscape Type

Quick observations:

- Lowland rural generaly dominates (but lesser contribution in later years)
- Spikes in some years for types (corresponding to special themes)
- Urban and Seascape both appear for first time in 1998; urban then constant presence, but seascape more variable until recent years 
```{r}
lspCounts <- yrdata %>%
  select(`Conference Year`,`Upland rural`, `Lowland rural`, Urban,	Riverscape, Seascape, `Undefined LspType`,Other) %>%
  mutate(yrsum = rowSums(.[2:8])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 


ggplot(lspCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(lspCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(lspCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

## Species

Quick observations:

- no clear patterns
- some years contain no Generic Habitat - is this real or a data entry issue?
```{r}
sppCounts <- yrdata %>%
  select(`Conference Year`,Mammals, Humans, Birds, Reptiles, Inverts, Plants, Amphibians, Fish, `Generic Habitat`,`Woodland Forests`) %>%
  mutate(yrsum = rowSums(.[2:11])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 

ggplot(sppCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(sppCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(sppCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

### Methods

Quick observations:

- empirical studies have decreased through time
- GIS and qualitative have increased through time
- Quantitative and theoretical quite steady through time (although theoretical does seem to have reduced after initial years)
```{r}
methodsCounts <- yrdata %>%
  select(`Conference Year`, Empirical, Theoretical, Qualitative, Quantitative, GIS, `Remote sensing`) %>%
  mutate(yrsum = rowSums(.[2:7])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 

ggplot(methodsCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(methodsCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(methodsCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

### Spatial Extent

Quick observations:

- no clear trends?
- Global studies only appear from 2014 onwards
```{r}
extentCounts <- yrdata %>%
  select(`Conference Year`, Micro, Mini, Local, Regional, National, Continental, Global,`Undefined Extent`) %>%
  mutate(yrsum = rowSums(.[2:9])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 

ggplot(extentCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(extentCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(extentCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```


### Temporal Extent

Quick observations:

- most studies have undefined temporal duration
- those that do are dominated by studies over decades and years 
```{r echo=F}
temporalCounts <- yrdata %>%
  select(`Conference Year`, Hours, Days, Weeks, Months, Years, Decades, Centuries, Longer, `Undefined Temporal`
) %>%
  mutate(yrsum = rowSums(.[2:10])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 

ggplot(temporalCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(temporalCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(temporalCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

### Concepts

Quick observations:

- Ecosystem services appear from 1998 and have grown recently
- climate change interactions have only become common recently (since 2008)
- 'Scale and scaling' and 'connectivity and fragmentation seem to have decreased oin recent years
- LUCC and Spatial Analysis are mainstays throughout
```{r}
conceptCounts <- yrdata %>%
  select(`Conference Year`, `PPS of landscapes`,
`Connectivity and fragmentation`, `Scale and scaling`,`Spatial analysis and modeling`,LUCC,`History and legacy`,`Climate change interactions`,`Ecosystem services`,`Landscape sustainability`,`Accuracy and uncertainty`
) %>%
  mutate(yrsum = rowSums(.[2:11])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 

ggplot(conceptCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(conceptCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(conceptCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

### Other Concepts

Quick observations:

- socio-economic studies have increased through time
- biodiversity has decreased through time
- Landscape management and Biodiversity peak in early 2000s
```{r}
othCCounts <- yrdata %>%
  select(`Conference Year`, `Green Infrastructure`,`Planning and Architecture`,`Management and Conservation`,`Cultural Landscapes`,`Socio-economic Dimensions`,Biodiversity,`Landscape Assessment`,`Catchment Based Approach`,`Invasives Pests Diseases`
) %>%
  mutate(yrsum = rowSums(.[2:10])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 

ggplot(othCCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(othCCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(othCCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```


# Analysis by Author Affiliation

```{r}
#spec(cpdata)

affildata <- cpdata %>%
  select_if(is.numeric) %>%
  gather(key = Affiliation, value = count, Academic:Private) %>%
  filter(count > 0) %>%
  group_by(`Affiliation`) %>%
  summarise_all(sum, na.rm=T) 
```

### Total Conference Contributions

Quick observations:

- Academic contributors dominate, followed by Government (but as shown above, Government contributions have decreased recently, replaced by NGOs)

```{r}

lspACounts <- affildata %>%
  select(`Affiliation`,`Upland rural`, `Lowland rural`, Urban,	Riverscape, Seascape, `Undefined LspType`,Other) %>%
  mutate(Asum = rowSums(.[2:8])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(lspACounts, aes(x=`Affiliation`, y=count)) + geom_bar(stat="identity")

```

## Landscape Type 

Stacked bar plots of contributions (by types and author affiliation)

### Using all landscape types

Quick observations:

- Business not good at reporting landscape type!
- Private have greatest proportions of Seascape and Other

```{r}

ggplot(lspACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(lspACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")
```

### Without 'Undefined LspType' and 'Other' landscape types

Quick observations:

- Government has greatest proportion of Upland Rural
- Business has greatest Urban proportion and smallest Lowland Rural proportion
- Academic dominates total number of all landscape types (with possible exception of Upland Rural)

```{r}

lspACounts <- affildata %>%
  select(`Affiliation`,`Upland rural`, `Lowland rural`, Urban,	Riverscape, Seascape) %>%
  mutate(Asum = rowSums(.[2:6])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(lspACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(lspACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```



## Species

Quick observations

- Academic seem to be majority by absolute number for all species
- Business and Private have greatest proportions of Generic Habitat
- NGOs have greatest proportion of Birds (RSPB?)

```{r}

speciesACounts <- affildata %>%
  select(`Affiliation`,Mammals, Humans, Birds, Reptiles, Inverts, Plants, Amphibians, Fish, `Generic Habitat`,`Woodland Forests`) %>%
  mutate(Asum = rowSums(.[2:11])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(speciesACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(speciesACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```


## Methods

Quick observations

- Academic seem to be majority by absolute number for all methods
- Business obviously lower proportion of empirical studies (expensive?), substituted by GIS and qualitative
- Government has smallest proportion of qualitative
- Private has greatest proprtion of theoretical, no RS and relatively little GIS (technical training?)


```{r}

methodsACounts <- affildata %>%
  select(`Affiliation`,Empirical, Theoretical, Qualitative, Quantitative, GIS, `Remote sensing`) %>%
  mutate(Asum = rowSums(.[2:7])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(methodsACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(methodsACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

## Spatial Extent

Quick observations

- Academic seem to be majority by absolute number for all extents
- Business have largest proportion of Global and National studies, with smallest proprtion of Local studies
- Private has larest proportion of Local and Mini studies (cost-related and given no RS and few GIS studies?)
- Academic: decreasing proportion Local -> Regional -> National -> Global
- Government: greater proportion of National than Regional

```{r}

spatialACounts <- affildata %>%
  select(`Affiliation`,Micro, Mini, Local, Regional, National, Continental, Global,`Undefined Extent`) %>%
  mutate(Asum = rowSums(.[2:9])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(spatialACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(spatialACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

## Temporal Extent

Quick observations

- Vast majority of all affiliations did not list temporal extent of the study
Academic seem to be majority by absolute number for all methods
- Not much more of interest here...

```{r}

temporalACounts <- affildata %>%
  select(`Affiliation`,Hours, Days, Weeks, Months, Years, Decades, Centuries, Longer, `Undefined Temporal`) %>%
  mutate(Asum = rowSums(.[2:10])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(temporalACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(temporalACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

## Concepts

Quick observations

- Academic seem to be majority by absolute number for all extents
- Business have greatest proportions of climate change and ecosystem services, less interested in history and legacy
- All other affiliations reasonably similar in terms of proportions

```{r}

conceptACounts <- affildata %>%
  select(`Affiliation`,`PPS of landscapes`,
`Connectivity and fragmentation`, `Scale and scaling`,`Spatial analysis and modeling`,LUCC,`History and legacy`,`Climate change interactions`,`Ecosystem services`,`Landscape sustainability`,`Accuracy and uncertainty`) %>%
  mutate(Asum = rowSums(.[2:11])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(conceptACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(conceptACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

## Other Concepts

Quick observations

- Academic seem to be majority by absolute number for all extents
- Government and NGOs have greater proportion of Management and Conservation than Academic
- Private low on biodiversity but higher on cultural landscapes, landscape assessment and planning

```{r}

oconceptACounts <- affildata %>%
  select(`Affiliation`,`Green Infrastructure`,`Planning and Architecture`,`Management and Conservation`,`Cultural Landscapes`,`Socio-economic Dimensions`,Biodiversity,`Landscape Assessment`,`Catchment Based Approach`,`Invasives Pests Diseases`) %>%
  mutate(Asum = rowSums(.[2:10])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(oconceptACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(oconceptACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

# Analysis by Landscape Type

```{r}
#spec(cpdata)

lspdata <- cpdata %>%
  select_if(is.numeric) %>%
  gather(key = LspType, value = count, `Upland rural`:Other) %>%
  filter(count > 0) %>%
  group_by(`LspType`) %>%
  summarise_all(sum, na.rm=T) 
```

Quick observations:

- Lowland rural dominate, followed by 'undefined' and Upland rural

```{r}

AlspCounts <- lspdata %>%
  select(LspType,Academic, Government,NGO,Business,Private) %>%
  mutate(Asum = rowSums(.[2:6])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -LspType, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(AlspCounts, aes(x=LspType, y=count)) + geom_bar(stat="identity")

```

## Author Affiliation

Quick observations:

- Academic are majority of all landscape types, with possible exception of Upland rural (Government?)
- 

```{r}
ggplot(AlspCounts, aes(x=LspType, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(AlspCounts, aes(x=LspType, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

## Species
Quick observations

- Animal types quite evenly distributed across Lowland rural
- Humans are large contributor to seascape studies (possibly by absolute number as well as relative)
- Generic habitat is large contributor across all landscape types

```{r}

specieslspCounts <- lspdata %>%
  select(LspType,Mammals, Humans, Birds, Reptiles, Inverts, Plants, Amphibians, Fish, `Generic Habitat`,`Woodland Forests`) %>%
  mutate(Asum = rowSums(.[2:11])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -LspType, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(specieslspCounts, aes(x=LspType, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(specieslspCounts, aes(x=LspType, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

## Methods

Quick observations

- 'Undefined landscape' studies are largely theoretical
- Lowland rural largely studies using empirical and quantitative methods
- Seascape studies have largest proportion of qualitative methods


```{r}

methodslspCounts <- lspdata %>%
  select(LspType,Empirical, Theoretical, Qualitative, Quantitative, GIS, `Remote sensing`) %>%
  mutate(Asum = rowSums(.[2:7])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -LspType, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(methodslspCounts, aes(x=LspType, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(methodslspCounts, aes(x=LspType, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

## Spatial Extent

Quick observations

- Urban landscape studies are dominated by Local scale analysis
- Upland rural have larger proportion of national studies than Lowland rural

```{r}

spatiallspCounts <- lspdata %>%
  select(LspType,Micro, Mini, Local, Regional, National, Continental, Global,`Undefined Extent`) %>%
  mutate(Asum = rowSums(.[2:9])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -LspType, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(spatiallspCounts, aes(x=LspType, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(spatiallspCounts, aes(x=LspType, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

```

## More here