Vectorize a Naive Python Loop into NumPy Ops, Before/After Speed Comparison

The problem, from first principles

Every operation in this curriculum's classical-ML and deep-learning tracks was implemented as a single vectorized NumPy expression, deliberately avoiding Python-level loops over individual elements — this question makes the reason for that convention explicit and measurable, rather than assumed.

From theory to code

Implement naive_dot_product(a, b) (a plain Python loop over two lists), vectorized_dot_product(a, b) (the identical computation as one NumPy call), and compare_speed(a, b), which runs both on the same data and reports both the results and the actual measured timings.

Constraints

• naive_dot_product: takes plain Python lists, returns a float.
• vectorized_dot_product: takes NumPy arrays, returns a float.
• compare_speed: takes NumPy arrays, returns a dict with naive_result, vectorized_result, naive_time, vectorized_time, speedup (naive_time / vectorized_time).
• compare_speed internally converts its array inputs to lists for naive_dot_product — it doesn't change what's being measured, only which representation each implementation actually expects.

Hints

Open one at a time. Each gives away a little more than the last.