Exercise 3: Reading the Illustrative Results — Possible Solution
====================================================================

WHAT MAE ≈ $2,100 TELLS YOU
------------------------------
Per this chapter's own definition, MAE is the average absolute error in
the target's original units. An MAE of about $2,100 means that,
typically, this model's predicted price lands within roughly $2,100 of
the car's real price — a plain, directly interpretable statement about
the model's own everyday prediction accuracy.

WHAT RMSE ≈ $2,800 TELLS YOU
------------------------------
Per this chapter, RMSE shares MAE's own units but is more sensitive to
large errors, since it's built from squared error. An RMSE of about
$2,800 — noticeably higher than the $2,100 MAE — indicates the model's
error profile isn't perfectly uniform; something is pulling the
squared-error-based number up higher than the plain-average-based
number would predict if every error were roughly the same size.

WHAT R² ≈ 0.82 TELLS YOU
------------------------------
Per this chapter, R² measures the proportion of the target's own total
variance the model accounts for. An R² of about 0.82 means the model
explains roughly 82% of the real variation in price across the
dataset — a strong, but not perfect, overall fit. The remaining ~18%
of price's own variance is not captured by mileage and year alone,
consistent with ml1-3's own honest acknowledgment that other,
unmeasured factors (like accident history) could still matter.

WHAT THE MAE-RMSE GAP SPECIFICALLY SUGGESTS
------------------------------
Per this chapter's own reasoning, "a large gap means a few big misses
are driving the error total; a small gap means errors are fairly
uniform in size." The gap here ($2,800 − $2,100 = $700, a meaningful
proportional difference) suggests the model's typical error is smaller
than $2,800 would imply on its own, but a handful of larger,
outsized misses are pulling RMSE noticeably higher than MAE — exactly
the signature this chapter's own Jaguar warn-box described: a small
number of unusual cases disproportionately inflating the squared-error-
based metric while barely moving the plain-average one.

WHY THESE THREE NUMBERS TOGETHER TELL A COHERENT STORY
------------------------------
Per this chapter's own closing summary, "a genuinely useful model, not a
perfect one, with a modest number of harder-to-predict cases worth a
closer look." MAE establishes the model is typically fairly accurate;
R² confirms it captures most (but not all) of the real variation in
price; the MAE-RMSE gap specifically flags that the remaining
inaccuracy isn't spread evenly — it's concentrated in a smaller number
of harder cases, which is a genuinely useful, actionable piece of
information a single metric alone wouldn't have revealed.

WHY THIS WORKS AS AN ANSWER
------------------------------
It interprets each of the three illustrative numbers individually using
the chapter's own definitions, and specifically explains what the gap
between MAE and RMSE suggests about the shape of the model's own error
distribution, tying it back to the chapter's own Jaguar-outlier
reasoning.