melt เพื่อเปลี่ยนเป็นรูปแบบยาว

When columns are really values in disguise, melt unpivots them into rows, turning a wide table into tidy long form. 🫠

Melt creates two new columns: one holding the old column names, one holding their values. Every cell becomes its own row.

Use id_vars to list columns that should stay put, like an id or name. Those repeat down the new long table.

Set value_vars to melt only specific columns. Anything left out and not in id_vars is simply dropped from the result.

The defaults are variable and value. Pass var_name and value_name to give them meaningful labels like month and sales.

Libraries like seaborn want long input. A quick melt lets you map one column to color or facet and plot many series at once.

Once data is long, summaries get simple. After melt you can groupby the new variable column and aggregate every series uniformly.

Melt and pivot undo each other. Go wide to long with melt, then pivot back, and you recover the original layout.

Melting multiplies rows: ten columns over five rows becomes fifty long rows. Expect the row count to grow fast.

For columns sharing a prefix like sales_2021, wide_to_long melts them while parsing the year out of each name.

A reliable pattern: name your id columns, name the value columns, and label the output. That gives a fully tidy long table.

Quick check on melting wide tables.

You used melt with id_vars, value_vars, and custom names to tidy wide data. Next: stack and unstack on a MultiIndex. 🎯