Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Soil metals example (R loader) #749

Draft
wants to merge 6 commits into
base: main
Choose a base branch
from
Draft
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 10 additions & 0 deletions examples/soil-metals/.gitignore
Original file line number Diff line number Diff line change
@@ -0,0 +1,10 @@
.DS_Store
dist/
docs/.observablehq/cache/
node_modules/
yarn-error.log
yarn.lock
.Rhistory
.Rapp.history
.RData
.Ruserdata
15 changes: 15 additions & 0 deletions examples/soil-metals/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,15 @@
# Soil metal analysis

This is an example Observable Framework project that explores soil metal concentrations in soils from mining districts in Moquegua, Peru. The purpose is to show an R data loader within a Framework project. Metals selected for [principal component analysis]((https://en.wikipedia.org/wiki/Principal_component_analysis)) are random; charts should not be interpreted as research findings.

Data source: Bedoya-Perales, N.S., Escobedo-Pacheco, E., Maus, D. et al. Dataset of metals and metalloids in food crops and soils sampled across the mining region of Moquegua in Peru. Sci Data 10, 483 (2023). https://doi.org/10.1038/s41597-023-02363-0

View the [live project](TODO).

## Data loader

A single R data loader (`soil-metals.zip.R`) reads in data from a local Excel file (`heavy-metals.xlsx`), does basic wrangling, then performs principal component analysis. Multiple outputs (as CSVs) are added to a Zip archive.

## Charts

All charts are created as functions in `index.md` using [Observable Plot](https://observablehq.com/plot/).
Binary file added examples/soil-metals/docs/data/heavy-metals.xlsx
Binary file not shown.
83 changes: 83 additions & 0 deletions examples/soil-metals/docs/data/soil-metals.zip.R
Original file line number Diff line number Diff line change
@@ -0,0 +1,83 @@
# Attach libraries (must be installed)
library(readxl)
library(dplyr)
library(tidyr)
library(janitor)
library(stringr)
library(readr)

# Access and wrangle data
# Get only soil metal contentrations
hm_soils <- read_xlsx("docs/data/heavy-metals.xlsx", skip = 1, na = "-") |>
clean_names() |>
slice(-1) |>
select(-c(3, 4, 5, 14, 15, 16, 19:49)) |>
rename(altitude_masl = altitude_m_a_s_l)

# Update column names to only metal (e.g. "aluminum")
metal_matrix <- c(sub("_.*", "", names(hm_soils)[13:35]))
colnames(hm_soils)[13:35] <- metal_matrix

# Replace values below detection limit with LOD / 2:
detect_function <- function(vec) {
ifelse(str_detect(vec, "<"),
as.numeric(str_replace(vec, "<", "")) / 2,
as.numeric(vec)
)
}

hm_clean <- hm_soils |>
mutate(across(13:35, detect_function))

# Principal component analysis
pca_metals <- c("aluminum", "arsenic", "barium", "calcium", "chromium", "cobalt", "copper", "iron", "magnesium", "lead", "potassium", "vanadium")

# Subset with location and sampling information, and metals above (complete)
soil_subset <- hm_clean |>
select(1:12, all_of(pca_metals)) |>
drop_na()

# Only metal variables for PCA
pca_subset <- soil_subset |>
select(-(1:12))

# PCA:
soil_pca <- prcomp(pca_subset, scale = TRUE)

# PCA results

# Loadings
var_loadings <- as.data.frame(soil_pca$rotation) |>
tibble::rownames_to_column(var = "metal")

# Variance explained
var_exp <- data.frame(
pc = names(var_loadings[2:length(var_loadings)]),
variance = soil_pca$sdev^2 / sum(soil_pca$sdev^2)
)

# Scores
obs_scores <- data.frame(soil_subset, soil_pca$x)

# Scaled loadings (as in factoextra)
scaling <- min(
(max(obs_scores["PC2"]) - min(obs_scores["PC2"]) / (max(var_loadings["PC2"]) - min(var_loadings["PC2"]))),
(max(obs_scores["PC1"]) - min(obs_scores["PC1"]) / (max(var_loadings["PC1"]) - min(var_loadings["PC1"])))
)

# Note: factoextra uses 0.7 * scaling
var_loadings_scaled <- var_loadings |>
mutate(
PC1_scale = 0.7 * scaling * PC1,
PC2_scale = 0.7 * scaling * PC2
)

# Add to zip archive, write to stdout
setwd(tempdir())

write_csv(hm_clean, "soil-metals.csv")
write_csv(var_loadings_scaled, "var-loadings.csv")
write_csv(var_exp, "var-explained.csv")
write_csv(obs_scores, "obs-scores.csv")

system("zip - -r .")
180 changes: 180 additions & 0 deletions examples/soil-metals/docs/index.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,180 @@
# Soil metal analysis
## In mining regions across Moquegua, Peru

```js

const soilMetals = FileAttachment("data/soil-metals/soil-metals.csv").csv({typed: true});

const varLoadings = FileAttachment("data/soil-metals/var-loadings.csv").csv({typed: true});

const varExplained = FileAttachment("data/soil-metals/var-explained.csv").csv({typed: true});

const obsScores = FileAttachment("data/soil-metals/obs-scores.csv").csv({typed: true});
```

```js
const pickMetal = Inputs.radio(varLoadings.map(d => d.metal), {label: "Pick metal:", value: "aluminum"});
const metal = Generators.input(pickMetal);
```


```js
function boxPlot(width, height) {
return Plot.plot({
width,
height: height - 100,
marginLeft: 100,
x: { grid: true, label: "Soil concentration (mg/kg)"},
y: { grid: true, label: "District", tickSize: 0 },
marks: [
Plot.boxX(soilMetals, {
x: metal,
y: "district",
fill: "gray",
opacity: 0.3,
stroke: null
}),
Plot.dot(soilMetals, {
x: metal,
y: "district",
fill: "darkgray",
opacity: 0.7
}),
Plot.tickX(
soilMetals,
Plot.groupY(
{ x: "median" },
{
x: metal,
y: "district",
stroke: "#ff725c",
strokeWidth: 4,
sort: { y: "x", reverse: true }
}
)
)
]
});
}
```

```js
const pickDistrict = Inputs.checkbox(
soilMetals.map((d) => d.district),
{
label: "Highlight district(s):",
unique: true,
value: ["Algarrobal", "Carumas", "Chojata", "Coalaque"],
sort: true
}
);

const inputDistrict = Generators.input(pickDistrict);
```


```js
function biplot(width, height) {
return Plot.plot({
width,
height: height - 120,
x: {label: `PC1 (${d3.format(".1%")(varExplained.map(d => d.variance)[0])})`, ticks: 6},
y: {label: `PC1 (${d3.format(".1%")(varExplained.map(d => d.variance)[1])})`, ticks: 6},
marks: [
Plot.frame({stroke: "#BCBCBC"}),
Plot.dot(obsScores, {
x: "PC1",
y: "PC2",
fill: "#BCBCBC",
opacity: 0.5,
r: 2.5
}),
Plot.hull(obsScores, {
x: "PC1",
y: "PC2",
fill: d => inputDistrict.includes(d.district) ? d.district : null,
opacity: 0.4
}),
Plot.hull(obsScores, {
x: "PC1",
y: "PC2",
stroke: d => inputDistrict.includes(d.district) ? d.district : null,
strokeWidth: 2
}),
Plot.dot(obsScores, {
x: "PC1",
y: "PC2",
filter: d => inputDistrict.includes(d.district),
fill: "district",
r: 3,
opacity: 0.8
}),
Plot.arrow(varLoadings, { x1: 0, y1: 0, x2: "PC1_scale", y2: "PC2_scale" }),
Plot.text(varLoadings, {
text: "metal",
x: "PC1_scale",
y: "PC2_scale",
dy: -5,
dx: 30,
fill: "black",
stroke: "white",
fontSize: 11
})
]
});
}
```

```js
function screeplot(width, height) {
return Plot.plot({
marginTop: 15,
width,
height,
x: {label: null},
y: {axis: null},
marks: [
Plot.barY(varExplained, {
x: "pc",
y: "variance",
fill: "gray",
sort: { x: "y", reverse: true }
}),
Plot.ruleY([0]),
Plot.text(varExplained, {
text: (d) => d3.format(".1%")(d.variance),
x: "pc",
y: "variance",
dy: -8
})
]
});
}
```

<div class="grid grid-cols-4" style="grid-auto-rows: 145px;">
<div class="card grid-colspan-2 grid-rowspan-4">
<h2>Principal component analysis</h2>
<h3>Axis scales are for observation scores. See table for loading values.</h3>
${pickDistrict}
${resize((width, height) => biplot(width, height))}
</div>
<div class="card grid-colspan-2 grid-rowspan-2">
<h2>Variance explained</h2>
${resize((width, height) => screeplot(width, height))}
</div>
<div class="card grid-colspan-2 grid-rowspan-2" style="padding: 0; border-radius: 12px; overflow: hidden;">
${Inputs.table(varLoadings.map(({PC1_scale, PC2_scale, ...rest}) => rest), {height: 320})}
</div>
</div>

<div class="grid grid-cols-4" style="grid-auto-rows: 160px;">
<div class="card grid-colspan-4 grid-rowspan-3">
<h2>Soil metal concentration by district</h2>
<h3>Dots are individual soil sample values; black line is the median for each district. Box is the interquartile range. All values in mg/kg.</h3>
${pickMetal}
${resize((width, height) => boxPlot(width, height))}
</div>
</div>

<div class="note" label="Data">Bedoya-Perales, N.S., Escobedo-Pacheco, E., Maus, D. et al. Dataset of metals and metalloids in food crops and soils sampled across the mining region of Moquegua in Peru. Sci Data 10, 483 (2023). <a href="https://doi.org/10.1038/s41597-023-02363-0">https://doi.org/10.1038/s41597-023-02363-0</a></div>
14 changes: 14 additions & 0 deletions examples/soil-metals/package.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,14 @@
{
"type": "module",
"scripts": {
"dev": "observable preview",
"build": "rm -rf dist && observable build",
"postinstall": "ln -sf ../../../../bin/observable-init.js node_modules/.bin/observable"
},
"dependencies": {
"@observablehq/framework": "link:../.."
},
"engines": {
"node": ">=20.6"
}
}