Group 2 — Risk factors
Studio wall
1 Spine
How do bodies carry risk, how do class and smoke talk, and where do children still die young?
This sentence stays at the top. Every panel must attach to it. If a plot cannot be introduced in a clause that hangs off this line, it is off the wall.
2 Introduction — the tables we own
This folder is ours for the ten days. Three tables, three shapes.
- Panel A (
panel-A.csv) —cardioRiskFactors: BMI, lipids, blood pressure, uric acid.sexandsmokhave already been recoded from integers to factors. - Panel B (
panel-B.csv) — Scottish Health Survey 1998: sex, social class, smoking, diagnosed CVD.areais a code, not a map. - Panel C (
panel-C.csv) — SOWC child / infant / neonatal mortality by country. Join by country name.
If a brick needs a histogram, it lives in A. If it needs a mosaic, it lives in B.
We load from this folder:
library(readr)
library(tidyverse)
A <- read_csv("panel-A.csv")
B <- read_csv("panel-B.csv")
C <- read_csv("panel-C.csv")Found tables train the mark. A campus Free Hunch, if we collect one, is a made panel of the same pair type — new people, old grammar. It does not replace A, B, or C.
3 Panel A — body / measure
bmi, choles, sys, dia, Uric are Quant. Do not average smok.
3.0.1 1. See the table
What is a case? Which columns are Quant / Qual? What is missing?
# glimpse() / inspect() / skim()3.0.2 2. Ask
One question this table can answer. One it cannot. Write both.
3.0.3 3. Choose a mark
Geom, map mark, or (later) test that matches the pair type. Name it before you code.
3.0.4 4. Predict
What do we think we will see? One sentence, before Run.
3.0.5 5. Make
Human brick first. Optional: one AI prompt, pasted below, only for what section 3 named.
# human chunk
# ## Prompt we actually used
# ai_draft
# ours3.0.6 6. Say
Caption + one sentence we will stand behind. Title is a sentence, not a column name.
3.0.7 7. Limit
What this panel cannot do. Which kit table we refused to abuse.
4 Panel B — category / talk
smoke × sex, cvd × sc (social class). area is forbidden as geography.
4.0.1 1. See the table
What is a case? Which columns are Quant / Qual? What is missing?
# glimpse() / inspect() / skim()4.0.2 2. Ask
One question this table can answer. One it cannot. Write both.
4.0.3 3. Choose a mark
Geom, map mark, or (later) test that matches the pair type. Name it before you code.
4.0.4 4. Predict
What do we think we will see? One sentence, before Run.
4.0.5 5. Make
Human brick first. Optional: one AI prompt, pasted below, only for what section 3 named.
# human chunk
# ## Prompt we actually used
# ai_draft
# ours4.0.6 6. Say
Caption + one sentence we will stand behind. Title is a sentence, not a column name.
4.0.7 7. Limit
What this panel cannot do. Which kit table we refused to abuse.
5 Panel C — place
under5_mortality_2018 and infant_mortality_2018 are rates. Join on country.
Geometry is not in the CSV. Join in R, then use the same seven headings.
# library(sf)
# library(rnaturalearth)
# world <- ne_countries(scale = "medium", returnclass = "sf")
# # join on country name; list the unmatched labels5.0.1 1. See the table
What is a case? Which columns are Quant / Qual? What is missing?
# glimpse() / inspect() / skim()5.0.2 2. Ask
One question this table can answer. One it cannot. Write both.
5.0.3 3. Choose a mark
Geom, map mark, or (later) test that matches the pair type. Name it before you code.
5.0.4 4. Predict
What do we think we will see? One sentence, before Run.
5.0.5 5. Make
Human brick first. Optional: one AI prompt, pasted below, only for what section 3 named.
# human chunk
# ## Prompt we actually used
# ai_draft
# ours5.0.6 6. Say
Caption + one sentence we will stand behind. Title is a sentence, not a column name.
5.0.7 7. Limit
What this panel cannot do. Which kit table we refused to abuse.
6 Conclusion
What the spine looks like after three panels.
- The question we will defend in ninety seconds
- What we will not claim
- Which lines were human, which were machine
- Limits: sample, missingness, the test we did not run
- Gift for the steal-this strip (palette, annotation, join trick)
# sessionInfo()