in_WorkedExample_Workings.Rmd

---
title: "Worked Example - Background workings"
subtitle: <h5 style="font-style:normal">GEOG-364 - Spatial Analysis</h4>
output: 
  html_document:
    toc: true
    toc_depth: 4
    toc_float: true
    theme: flatly
---


<style>
p.comment {
background-color: #DBDBDB;
padding: 10px;
border: 1px solid black;
margin-left: 0px;
border-radius: 5px;
font-style: normal;
}

h1.title {
  font-weight: bold;
  font-family: Arial;  
}

h2.title {
  font-family: Arial;  
}

</style>


<style type="text/css">
 TOC {
  font-size: 12px;
  font-family: Arial;
}
</style>

\
```{r setup, include=FALSE,echo=FALSE}
knitr::opts_chunk$set(echo = TRUE, warning=FALSE, message = FALSE,dpi = 300, dev = "svg")

# invisible data read
library(tidyverse)
library(sp)
library(sf)
library(readxl)
library(skimr)
library(tmap)
library(viridis)
library(usmap)
library(rnaturalearth)
library(ggpubr)
library(gridExtra)
library(ceramic)
library(leaflet)
library(leaflet.extras)
library(leaflet.providers)
library(raster)
library(ggmap)
library(dismo)
library(maps)


as.Date_origin <- function(x){
  format.Date(as.Date(x, origin = '1970-01-01'), format = '%b-%d')
}

Sys.setenv(MAPBOX_API_KEY="pk.eyJ1IjoiaGdyZWF0cmV4IiwiYSI6ImNrbmdpNDN6djBkZTQydXBhNDc1enJjZnkifQ.xQWYvOFK5BKQoNWoOacDgA")


```


<style>
p.comment {
background-color: #DBDBDB;
padding: 10px;
border: 1px solid black;
margin-left: 0px;
border-radius: 5px;
font-style: normal;
}

h1.title {
  font-weight: bold;
  font-family: Arial;  
}

h2.title {
  font-family: Arial;  
}

</style>


<style type="text/css">
#TOC {
  font-size: 11px;
  font-family: Arial;
}
</style>

# What is this?

In part 2 (to be uploaded), I show my actual example report analysing farmers markets in Iowa. So you can see what I would actually submit in the next section.

These are my background workings - all my text are things I might think in my head, you don't need to write all of this.  But I hope you can follow this and do something similar for your own work.

<br><br>


# 1. Set up and libraries

My example tutorial is set in Iowa on farmers markets.  First, set up libraries

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

library(sp)
library(sf)
library(tidyverse)
library(tmap)
library(readxl)
library(skimr)
library(spatstat)

```


<br><br>


# 2. Read in the data

Now I will read in my data, on farmers markets in Iowa.  Sometimes adding in the na option helps R recognise missing values, so I will try that.


```{r}
IA_market   <- readxl::read_excel("./Data/Farmers_Markets_Iowa.xlsx",na="NA")
```


```{r}
names(IA_market)
```

OK, I can see I have the columns above and when I click on the IA_market name in the Environment tab I can see more details. First, I'll take a quick look at the summary.

```{r}
summary(IA_market)
```

<br><br>


# 3. Initial quality control

From the summary above, I can see some weird issues.   

## 3.1. Problem with X

Hmm there is a very small longitude value in X that can't exist

```{r}
# get rid of the row with the weird value
# I could also set it to NA
IA_market <- IA_market[which(IA_market$X > -990),]

# and print the summary
summary(IA_market[,c("X","Y")])


```

OK that looks better.

<br><br>


## 2. Remove all the rows with missing locations

I know I don't need any row with missing location data, so let's remove

```{r}
IA_market <- IA_market[complete.cases(IA_market$X), ]

IA_market <- IA_market[complete.cases(IA_market$Y), ]

```

<br><br>

# 4. Choose Columns

I don't need all this data, so let's keep things neat.

I quite like the glimpse function from tidyverse for this. Here are my columns (same as name command)

```{r}
glimpse(IA_market)
```

 - I know from google that FID is a federal identifier that might be useful, so I'll keep that.  
 - Looking at the data, "location" is the full address, so let's remove that.
 - All the data is in Iowa, so let's remove State
 - Open_Hours and Open Dates feels too complicated, so i'm removing them
 
```{r}
IA_market <- IA_market[,c("X","Y","FID",
                                        "City","County",
                                        "Market_Name",
                                        "Weekday")]
summary(IA_market)

```

<br><br>

# 5. Create families of grouped data

Some of my data like market name is going to be unique to each market. Others like weekday could be considered data groups/families. In R this is called a factor.

To know the difference, I am thinking about whether it is useful to make a summary table of counts for that value

So 

```{r}
IA_market$City <- as.factor(IA_market$City)
IA_market$County <- as.factor(IA_market$County)
IA_market$Weekday <- as.factor(IA_market$Weekday)

```

Now when I summarise, I can immediately see that there are 12 markets on Fridays.

```{r}
summary(IA_market)
```

<br><br>


# 6. Rename columns

I know my X and Y data are really latitude and longitude, so I will rename them. to make sure I don't mess up, lots of printing out

```{r}
names(IA_market)[1] 
names(IA_market)[1] <- "Longitude" 
names(IA_market)[1] 

names(IA_market)[2] 
names(IA_market)[2] <- "Latitude" 
names(IA_market)[2] 

summary(IA_market)
```

<br><br>


# 7. Make a spatial version

OK I know my data is in Lat/Long, so when I make it spatial, I include the crs code 4326 (yours is almost certainly likely to be the same)

```{r}
# I know that it was in lat/long originally, hence the crs
IA_market.sf <- st_as_sf(IA_market,
                         coords=c("Longitude","Latitude"),
                         crs=4326)


# make a quick plot
tmap_mode("view")
qtm(st_geometry(IA_market.sf))
```

<br><br>


# 8. More quality control 

You can see in the plot above that there appears to be one point that is not in Iowa. So now I will look at the long/lat columns. I can also filter my data in the View tab.

Note i'm doing this on the original data not the sf.

```{r}
summary(IA_market[,1:2])
hist(IA_market$Longitude)

```

I feel that maybe that point at Long \~ -88 is wrong. Let's take a look.

```{r}

IA_market[IA_market$Longitude > -90,]

```

I'm guessing there's a typo - so I could just google this and fix it! but for now, let's remove and try again.

```{r}
# choose all the other ones, note the <= instead of >
IA_market <- IA_market[which(IA_market$Longitude < -90),]

# RECREATE THE SPATIAL DATA 
IA_market.sf <- st_as_sf(IA_market,coords=c("Longitude","Latitude"),
                         crs=4326)

# make a quick plot
tmap_mode("view")
qtm(IA_market.sf)
```

Much better.  Now let's change to a local map projection of my choice.I am choosing UTM Iowa from Tutorial 9.

```{r}
# And change to a map projection of your choice. 
IA_market.sf.utm <- st_transform(IA_market.sf,3744)

# make a quick plot
tmap_mode("view")
qtm(st_geometry(IA_market.sf.utm))

```

and let's make it a ppp variable using spatstat. Looking good!

```{r}
IA_market.ppp <- as.ppp(IA_market.sf.utm)
plot(IA_market.ppp)
```