Shape وSize وdtype

Every array carries metadata about itself. Three labels tell you almost everything: its shape, its size, and its dtype. Let's read each one. 🔍

The shape is a tuple of dimension lengths. A 3-by-4 grid reports (3, 4): three rows down, four columns across.

Shape is your first sanity check. If a model expects (100, 5) and you hand it (5, 100), the shape mismatch is where bugs hide.

The ndim attribute counts dimensions. A flat row is 1, a table is 2, a stack of tables is 3. It is just the length of the shape.

The size is the total number of elements, the product of every shape value. A (3, 4) array has a size of 12.

The dtype names the single type all elements share, like int64 or float64. It controls both precision and memory use.

You can set the type yourself with the dtype argument. Use a smaller type like int32 to save memory on huge arrays.

float32 holds fewer decimals than float64. Tiny rounding gaps add up, so pick the type your numbers actually need.

Already have an array? astype returns a copy in a new type, handy for turning floats into clean integers.

The itemsize tells you bytes per element. Multiply it by size to estimate the array's total memory footprint.

Checking shape and dtype first saves hours. They reveal mismatched data and wrong types before a calculation quietly goes wrong. ✅

Let's test your read of an array's shape.

You can now read an array at a glance: shape for layout, ndim and size for counts, and dtype for the shared type. These four checks catch most bugs early. 🎉