Public
Edited
Nov 30, 2022
1 fork
BRATENAHL
Questions we want to ask:
Are some officers being more racist than others?
Missing data: Out of ${totalStops} stops, ${blankRace} had missing value for race, and ${unknownRace} had race unknown. That's ${parseFloat((blankRace + unknownRace)/totalStops *100).toFixed(2)}% of total cases.
Inputs.table(dt)
totalStops = dt.count().objects()[0]["count"]
blankRace = dt
.filter((d) => d["race"] == "")
.count()
.objects()[0]["count"]
unknownRace = dt
.filter((d) => d["race"] == "U")
.count()
.objects()[0]["count"]
stopsbyrace = dt
.groupby("race")
.count()
.derive({
pct: (row) => (row.count / aq.op.sum(row.count)) * 100
})
Inputs.table(stopsbyrace)
racenotlisted = dt
.filter((d) => d["race"] === "")
.groupby("dispo")
.count()
.derive({
pct: (row) => (row.count / aq.op.sum(row.count)) * 100
})
.view()
raceGrouped = dt
.groupby("race", "dispo")
.count()
.groupby("race")
.derive({
pct: (row) => row.count / aq.op.sum(row.count)
})
.orderby("race", aq.desc("pct"))
.derive({
count: aq.escape((d) => (typeof d.count == "undefined" ? 0 : d.count))
})
.pivot("dispo", "count")
Inputs.table(raceGrouped)
racearrests = dt
.filter((d) => d["dispo"] === "ARREST")
.groupby("race")
.count()
.derive({
pct: (row) => (row.count / aq.op.sum(row.count)) * 100
})
.orderby("race", aq.desc("pct"))
Inputs.table(racearrests)
racearrestspct = dt
.filter((d) => d["dispo"] === "ARREST")
.filter((d) => d["race"] !== "")
.groupby("race")
.count()
.derive({
pct: (row) => (row.count / aq.op.sum(row.count)) * 100
})
.orderby("race", aq.desc("pct"))
racecitation = dt
.filter((d) => d["dispo"] === "CITE ISSUED")
.groupby("race")
.count()
.derive({
pct: (row) => (row.count / aq.op.sum(row.count)) * 100
})
.orderby("race", aq.desc("pct"))
Inputs.table(racecitation)
racecitationpct = dt
.filter((d) => d["dispo"] === "CITE ISSUED")
.filter((d) => d["race"] !== "")
.filter((d) => d["race"] !== "U")
.groupby("race")
.count()
.derive({
pct: (row) => (row.count / aq.op.sum(row.count)) * 100
})
.orderby("race", aq.desc("pct"))
dt
.filter((d) => d.race == "A")
.derive({
consequence: (d) =>
op.includes(["ARREST", "CITE ISSUED", "CITE SRAD"], d["dispo"])
? "consequence"
: "no_consequence"
})
.orderby((d) => d.consequence)
.view()
dt
.derive({
consequence: (d) =>
op.includes(["ARREST", "CITE ISSUED", "CITE SRAD"], d["dispo"])
? "consequence"
: "no_consequence"
})
.groupby("race", "consequence")
.count()
.impute({ count: () => 0 })
.groupby("race")
.derive({
pct: (row) => row.count / aq.op.sum(row.count)
})
.orderby("race", aq.desc("pct"))
.derive({
count: aq.escape((d) => (typeof d.count == "undefined" ? 0 : d.count))
})
.pivot("consequence", "count")
.derive({
pct: (row) =>
Math.round(
(row.consequence / (row.consequence + row.no_consequence)) * 10000
) / 100
})
.view()
// Parse datetime objects and extract day of week
dtDayOfWeek = dt
.derive({ dateWithoutTime: (d) => op.replace(d["date"], " 0:00", "") })
.derive({ dateSplit: (d) => op.split(d["dateWithoutTime"], "/") })
.spread("dateSplit", { as: ["monthClean", "dateClean", "yearClean"] })
.derive({
dateTimeObject: (d) =>
// Really weird how op.datetime counts month from 0. Just, why?
op.datetime(d["yearClean"], d["monthClean"] - 1, d["dateClean"])
})
.derive({
dayOfWeekClean: aq.escape((d) =>
d["dateTimeObject"].toLocaleString("en-us", { weekday: "long" })
)
})
stops2020 = dtDayOfWeek.filter((d) => d.yearClean === "2020")
totalStops2020 = stops2020.count().objects()[0]["count"]
stopsbyrace20 = stops2020
.groupby("race")
.count()
// .groupby("race")
.derive({
pct: (row) => row.count / aq.op.sum(row.count)
})
.orderby("race", aq.desc("pct"))
Inputs.table(stopsbyrace20)
stops2021 = dtDayOfWeek.filter((d) => d.yearClean === "2021")
totalStops2021 = stops2021.count().objects()[0]["count"]
stopsbyrace21 = stops2021
.groupby("race")
.count()
// .groupby("race")
.derive({
pct: (row) => row.count / aq.op.sum(row.count)
})
.orderby("race", aq.desc("pct"))
Inputs.table(stopsbyrace21)
racedispoGrouped21 = stops2021
.groupby("race", "dispo")
.count()
.groupby("race")
.derive({
pct: (row) => row.count / aq.op.sum(row.count)
})
.orderby("race", aq.desc("pct"))
.derive({
count: aq.escape((d) => (typeof d.count == "undefined" ? 0 : d.count))
})
.pivot("dispo", "count")
bybadge = dt.groupby("badge1", "race").count()
Inputs.table(bybadge)
racebybadge = bybadge
.groupby("badge1")
.pivot("race", "count")
.rename(aq.names("badge1", "unlisted"))
.derive({
// You'd need to repeat this for all the columns you want to convert
// For a reason I'm not sure of yet, aq.escape() is needed, otherwise the `typeof` causes problems.
B: aq.escape((d) => (typeof d.B == "undefined" ? null : d.B)),
F: aq.escape((d) => (typeof d.F == "undefined" ? null : d.F)),
A: aq.escape((d) => (typeof d.A == "undefined" ? null : d.A)),
U: aq.escape((d) => (typeof d.U == "undefined" ? null : d.U)),
I: aq.escape((d) => (typeof d.I == "undefined" ? null : d.I)),
W: aq.escape((d) => (typeof d.W == "undefined" ? null : d.W)),
U: aq.escape((d) => (typeof d.U == "undefined" ? null : d.U)),
unlisted: aq.escape((d) =>
typeof d.unlisted == "undefined" ? null : d.unlisted
)
})
.derive({
total: (row) => row.unlisted + row.A + row.W + row.B + row.F + row.I + row.U
})
.derive({
pctblack: (row) =>
(100 * row.B) / (row.A + row.W + row.B + row.F + row.I + row.U)
})
.derive({
pctblank: (row) =>
(100 * row.unlisted) /
(row.A + row.W + row.B + row.F + row.I + row.U + row.unlisted)
})
.filter((d) => d["total"] > 10)
Inputs.table(racebybadge)
Inputs.table(dtDayOfWeek)
stops
racebybadge.view()
Plot.plot({
inset: 8,
grid: true,
color: {
legend: true
},
marks: [
Plot.dot(racebybadge, { x: "pctblack", y: "pctblank", r: "total" }),
Plot.linearRegressionY(racebybadge, {
x: "pctblack",
y: "pctblank",
stroke: "steelblue",
ci: 0.95
})
]
})
locationsbyrace = toplocations
.groupby("loc1", "loc2")
.derive({
race: aq.escape((d) => (typeof d.race == "undefined" ? null : d.race))
})
.pivot("race", "count")
.rename(aq.names("loc1", "loc2", "unlisted"))
.derive({
// You'd need to repeat this for all the columns you want to convert
// For a reason I'm not sure of yet, aq.escape() is needed, otherwise the `typeof` causes problems.
B: aq.escape((d) => (typeof d.B == "undefined" ? null : d.B)),
F: aq.escape((d) => (typeof d.F == "undefined" ? null : d.F)),
A: aq.escape((d) => (typeof d.A == "undefined" ? null : d.A)),
U: aq.escape((d) => (typeof d.U == "undefined" ? null : d.U)),
I: aq.escape((d) => (typeof d.I == "undefined" ? null : d.I)),
W: aq.escape((d) => (typeof d.W == "undefined" ? null : d.W)),
U: aq.escape((d) => (typeof d.U == "undefined" ? null : d.U)),
unlisted: aq.escape((d) =>
typeof d.unlisted == "undefined" ? null : d.unlisted
)
})
.derive({
total: (row) => row.unlisted + row.A + row.W + row.B + row.F + row.I + row.U
})
.filter((d) => d["total"] > 10)
.derive({
pctblack: (row) =>
(100 * row.B) / (row.A + row.W + row.B + row.F + row.I + row.U)
})
Inputs.table(toplocations)
Inputs.table(locationsbyrace)
addTooltips(
Plot.plot({
width: 1000,
overflow: "scroll",
color: {
type: "linear",
range: ["steelblue", "orange"] // uses d3.interpolateRgb
},
x: {
domain: d3.sort(locationsbyrace, (d) => -d.total).map((d) => d.loc1)
},
y: {
grid: true
},
marks: [
Plot.barY(locationsbyrace, {
x: "loc1",
y: "total",
fill: (d) => -d.pctblack,
title: (d) => d.loc1 + " & " + d.loc2 + "\n" + d.pctblack + "% Black"
}),
Plot.ruleY([0])
]
})
)
Inputs.table(racedispoGrouped21)
raceGrouped
.derive({
consequences: (d) => d.ARREST + d["CITE ISSUED"] + d["CITE SRAD"]
})
.view()
raceGrouped.view()
Inputs.table(racearrestspct)
Inputs.table(racecitationpct)
Inputs.table(racenotlisted)
stops = aq.from(
await FileAttachment("Jan 1 2020- Sept 15 2022 .xls - callracesex.csv").csv()
)
jan12020Sept152022XlsCallracesex = FileAttachment("Jan 1 2020- Sept 15 2022 .xls - callracesex.csv").csv()
import { addTooltips } from "@mkfreeman/plot-tooltip"
import { Plot } from "@mkfreeman/plot-tooltip"
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.