05-ialeUK25-analysisMethods.Rmd

# Analysis by Methods

Bar charts to examine how contributions to conferences vary by methods

```{r}
#spec(cpdata)

metdata <- cpdata %>%
  select_if(is.numeric) %>%
  gather(key = Method, value = count, Empirical:`Remote sensing`) %>%
  filter(count > 0) %>%
  group_by(`Method`) %>%
  summarise_all(sum, na.rm=T) 
```

## Total Conference Contributions

Quick observations:

- Empirical and Quantitative studies dominate
- little remote sensing

```{r}

metdata %>%
  select(Method, count) %>%
  mutate(prop = count/sum(count)) %>%
  mutate(prop = round(prop,3)) %>% 
  kable() %>%
  kable_styling() %>%
  scroll_box(width = "100%")
    

ggplot(metdata, aes(x=Method, y=count)) + 
  geom_bar(stat="identity") + 
  geom_text(aes(x=Method, y=max(count), label = paste0(round(100*count / sum(count),1), "%"), vjust=-0.25))


```

## Author Affiliation

Quick observations:

- pretty consistent distribution of methods across affiliations

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


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

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

```

## Landscape Type 

### Using all landscape types

Quick observations: 

- Qualitative studies are most evenly distributed across landscape types
- Theoretical studies most commonly do not define a landscape type

```{r}
lspCounts <- metdata %>%
  select(Method,`Upland rural`, `Lowland rural`, Urban,	Riverscape, Seascape, `Undefined LspType`,Other) %>%
  mutate(sum = rowSums(.[2:8])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -Method, -sum) %>%
  mutate(prop = count / sum)  #calculate proportion

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

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

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

Quick observations:

- This is generally less informative than with 'un-defined' (in contrast to other classifications)
- Does highlight that qualitative studies are most prevalent in seascape studies

```{r}

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

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

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

```

## Species

Quick observations

- Qualitative studies most commonly study humans

```{r}

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

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

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

```


## Spatial Extent

Quick observations:

- Theoretical studies have largely undefined extent
- No micro qualitative studies
- very few global empirical studies, and no GIS global studies
```{r}
spatialCounts <- metdata %>%
  select(Method, Micro, Mini, Local, Regional, National, Continental, Global,`Undefined Extent`) %>%
  mutate(sum = rowSums(.[2:9])) %>%
  gather(key = Type, value = count, -Method, -sum) %>%
  mutate(prop = count / sum) 

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

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

```

## Temporal Extent 

### With undefined

Quick observations:

- Difficult to see much; examine without 'undefined' 
```{r echo=F}
temporalCounts <- metdata %>%
  select(Method, Hours, Days, Weeks, Months, Years, Decades, Centuries, Longer, `Undefined Temporal`
) %>%
  mutate(sum = rowSums(.[2:10])) %>%
  gather(key = Type, value = count, -Method, -sum) %>%
  mutate(prop = count / sum) 

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

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

```

### Without Undefined

- Theoretical studies are years or longer 
- Actually, vast majority of studies are months or longer (should see this when splitting by temporal extent)

```{r echo=F}
temporalCounts <- metdata %>%
  select(Method, Hours, Days, Weeks, Months, Years, Decades, Centuries, Longer 
) %>%
  mutate(sum = rowSums(.[2:9])) %>%
  gather(key = Type, value = count, -Method, -sum) %>%
  mutate(prop = count / sum) 

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

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

```

## Concepts

Quick observations:

- Surprisingly few theoretical studies of scale and scaling?
- Relatively large number of qualitative studies are about landscape sustainability
- Relatively few empirical studies of ecosystem services
- GIS studies have largest proportion of spatial analysis and modellng (unsurprisingly)

```{r}
conceptCounts <- metdata %>%
  select(Method, `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(sum = rowSums(.[2:11])) %>%
  gather(key = Type, value = count, -Method, -sum) %>%
  mutate(prop = count / sum) 

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

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

```


## Other Concepts

Quick observations:

- Qualitative studies have relatively low proportion of biodiversity studies, but have relatively large proportion of socio-economic and cultural
- Theoretical studies have similar distribution to qualitative slightly more biodiversity at expense of cultural

```{r}
othCCounts <- metdata %>%
  select(Method, `Green Infrastructure`,`Planning and Architecture`,`Management and Conservation`,`Cultural Landscapes`,`Socio-economic Dimensions`,Biodiversity,`Landscape Assessment`,`Catchment Based Approach`,`Invasives Pests Diseases`
) %>%
  mutate(sum = rowSums(.[2:10])) %>%
  gather(key = Type, value = count, -Method, -sum) %>%
  mutate(prop = count / sum) 

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

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

```