25 Essential pandas Tricks for Data Manipulation in Python

This article shares 25 useful pandas tricks for Python data analysis, illustrating how to inspect the pandas version and its dependencies with pd.__version__ and pd.show_versions() .

It demonstrates creating example DataFrames from a dictionary and from numpy.random.rand() , and how to assign custom column names.

Column renaming is covered using df.rename() , direct assignment to df.columns , and df.columns.str.replace() to replace spaces with underscores.

Adding prefixes or suffixes to column names is shown with df.add_prefix() and df.add_suffix() .

Row order can be reversed with df.loc[::-1] and reset to a default integer index using df.loc[::-1].reset_index(drop=True) .

Column order reversal is achieved via df.loc[:, ::-1] , which swaps the column positions.

Selecting columns by data type is illustrated with df.select_dtypes(include='number') for numeric columns and similar calls for object types.

Converting object columns to numeric types uses df[col].astype() , pd.to_numeric() with errors='coerce' , and filling missing values with df.fillna(0) or applying the conversion across the DataFrame with df.apply(pd.to_numeric, errors='coerce') .

Memory usage can be reduced by reading only needed columns via the usecols parameter and converting suitable object columns to the category dtype, shrinking a DataFrame from 30.4 KB to 2.3 KB.

For datasets split across multiple CSV files, the article shows how to concatenate them row‑wise using glob.glob() , sorted() , and pd.concat(..., ignore_index=True) , as well as column‑wise concatenation with pd.concat(..., axis=1) .