Exercise 1: Matching Three Techniques to Their EDA Step — Possible Solution
====================================================================

df.describe() → STEP 2, SUMMARY STATISTICS
------------------------------
Per this chapter's own step list, Step 2 is "Summary statistics.
df.describe() — every one of ds1-6's own vocabulary terms (mean, std,
Q1, median, Q3) for every numeric column, in one call." This matches
directly and by name — df.describe() is the chapter's own example for
this exact step.

A BOX PLOT OF price GROUPED BY fuel_type → STEP 4, BIVARIATE ANALYSIS
------------------------------
Per this chapter, Step 4 is "Bivariate analysis — two variables at a
time... a box plot (ds1-8) of price grouped by fuel_type." A box plot
of price grouped by fuel_type involves exactly two variables — the
numeric variable being summarized (price) and the categorical variable
used to split it into groups (fuel_type) — which is precisely why this
belongs to bivariate analysis rather than univariate (which would only
look at price OR fuel_type alone, not their relationship) or
multivariate (which would need three or more variables involved at
once).

A CORRELATION HEATMAP ACROSS year/mileage/price → STEP 5, MULTIVARIATE ANALYSIS
------------------------------
Per this chapter, Step 5 is "Multivariate analysis — three or more
variables at once. A correlation heatmap and a pair plot (ds1-8) across
every numeric column simultaneously." A heatmap covering year, mileage,
and price involves three separate numeric columns examined together in
a single chart, computing every pairwise correlation among all three at
once — exactly the "three or more variables at once" threshold this
chapter uses to distinguish multivariate analysis from the two-variable
scope of bivariate analysis.

WHY THE STEP BOUNDARIES ARE DEFINED BY VARIABLE COUNT
------------------------------
Across all three examples, the deciding factor is simply how many
variables are involved in the technique at once: one (univariate), two
(bivariate), or three-plus (multivariate) — a consistent, mechanical
way to classify any given chart or statistic into its correct step,
independent of which specific chart type or statistic is being used.

WHY THIS WORKS AS AN ANSWER
------------------------------
It matches each of the three named techniques to its step using the
chapter's own explicit step descriptions and worked examples, and
explains the underlying variable-count rule (one/two/three-or-more)
that determines which step any given technique belongs to.