ปรับรูปร่างและทำอาร์เรย์ให้แบน

An array is just numbers plus a shape. Reshaping rearranges those numbers into new dimensions without copying or changing a single value.

Use reshape to lay 12 numbers out as a 3 by 4 grid. The total count must stay the same, or NumPy raises an error.

The new shape only works if the dimensions multiply to the original size. Here 3 times 4 equals 12, so the reshape fits perfectly.

Pass -1 for one dimension and NumPy computes it for you. It is a handy shortcut when you only know one side of the grid.

Need a flat line of numbers again? ravel collapses any array into a single dimension, reading values row by row.

The flatten method also returns 1D, but always as a fresh copy. Editing the result never touches the original array.

Unlike flatten, ravel often returns a view that shares memory with the source. Change the view and the original can change too.

A 1D array can grow into a column with reshape. Turning shape (3,) into (3, 1) is common before stacking or matrix math.

The .T attribute swaps rows and columns. A 2 by 3 array instantly becomes 3 by 2, mirroring the data across its diagonal.

You cannot reshape 10 numbers into a 3 by 4 grid. The element count on both sides must match, so plan dimensions before you reshape.

Models and plotting tools expect specific shapes. Reshaping lets one block of numbers serve a vector, a matrix, or an image as needed.

You want a guaranteed independent 1D copy of an array.

You reshaped data while keeping its values, used -1 for auto sizing, and learned that flatten copies while ravel may share memory. 🎯