Digital Humanist. Manhattan-based.
Published unlisted
Edited
Jun 11, 2019
colors = d3.scaleOrdinal(d3.schemeAccent)
colors.domain()
viewof CLOSENESS_THRESHOLD = slider({'range': [0, 1], 'value': 0.2})
viewof STEP = slider({'min': 1, 'max': 50, 'value': 10})
FIXED_ITEM = 1
margin = ({top: 30, right: 80, bottom: 30, left: 30});
width = 700 - margin.left - margin.right
height = 1000 - margin.top - margin.bottom
sim_mat
embed({data: {values: sim_mat}, "mark": "rect",
"mark": "line",
"encoding": {
"y": {
"field": "sim",
"type": "quantitative"
},
"x": {
"field": "rank",
"type": "quantitative"
},
"row": {"field": "k2"},
"column": {"field": "k1"},
"detail": {"field": "interaction"},
"color": {"field": "k1"}
},"transform": [
{
"filter": "datum.k1 != datum.k2"
},
{
"as": "interaction", "calculate": "'' + datum.k1 + datum.k2"
},
{
"groupby": ["k1", "k2"],
"window": [{
"op": "rank",
"as": "rank",
}],
"sort": [{ "field": "sim", "order": "descending" }]
}, {
"filter": `datum.rank <= ${Math.sqrt(sim_mat.length) * 4 / Math.sqrt(ids.length)}`
}
],
"width":400/ids.length, "height": 400/ids.length, "title": `Windowed similarity`})
metrics = new Object({"jaccard": jaccard, "cosine": cosine})
cosine = function(a, b) {
// Cosine distance between two key-value arrays.
return d3.sum(Object.keys(a).map(k => {
const v = a[k];
return b[k] ? b[k] * v : 0
}))/
Math.sqrt(d3.sum(Object.values(a).map(d => d*d)))/
Math.sqrt(d3.sum(Object.values(b).map(d => d*d)))
}
import { Library } from '@bmschmidt/javascript-bindings-to-the-hathi-features-data'
import { jaccard, similarity_matrix } from '@bmschmidt/book-visualizations-sandbox'
library = new Library().fetch_all(ids)
function full_similarity_matrix(library) {
const { m } = library;
const books = Array.from(m.values())
function label(book) {
const { data } = book;
return `${data.metadata.pubDate}-${data.metadata.title}-${book.id}`
}
books.sort((a, b) => (label(a) > label(b)))
let data = []
// Compare every book to every other book.
books.forEach((first, i) => {
console.log(`Building sims for ${first.id}`)
books.forEach((second, j) => {
const prefs = {'doc': 'chunk', 'size': chunk_size}
if (i <= j) {
const aa = first.count(prefs)//.filter(d => Math.random() < (1 - drop_share))
const bb = second.count(prefs)//.filter(d => Math.random() < (1 - drop_share))
const dists = similarity_matrix(aa, bb, label(first), label(second))
data = data.concat(dists)
}
})
})
return data
}
sim_mat = full_similarity_matrix(library)
vegalite = require("@observablehq/vega-lite@0.1")
embed = require('vega-embed')
d3Fetch = require('d3-fetch')
d3 = require('d3@5')
import {slider, select} from "@jashkenas/inputs"
One platform to build and deploy the best data apps
Experiment and prototype by building visualizations in live JavaScript notebooks. Collaborate with your team and decide which concepts to build out.
Use Observable Framework to build data apps locally. Use data loaders to build in any language or library, including Python, SQL, and R.
Seamlessly deploy to Observable. Test before you ship, use automatic deploy-on-commit, and ensure your projects are always up-to-date.