10 Pandas One-Liners Every Data Analyst Should Know

Pandas can either feel like magic or like fighting NumPy with both hands tied. Ten one-liners cover the vast majority of day-to-day analyst work — memorise these and your notebooks shrink by half.

The ten

• Top rows per group — df.groupby('city').head(3).
• Conditional column — df['tier'] = np.where(df['amount'] > 1000, 'high', 'low').
• Cross-tab — pd.crosstab(df.city, df.product, values=df.revenue, aggfunc='sum').
• Rolling average — df['ma7'] = df['sales'].rolling(7).mean().
• Percent change — df['pct'] = df['sales'].pct_change().
• Pivot — df.pivot_table(index='date', columns='city', values='sales', aggfunc='sum').
• Drop duplicates by column — df.drop_duplicates(subset=['email'], keep='last').
• Filter by multiple values — df[df['city'].isin(['Chennai', 'Pune'])].
• String contains — df[df['title'].str.contains('Engineer', case=False, na=False)].
• Apply with axis — df.apply(lambda r: r['a'] + r['b'], axis=1).

The bonus skill

Learn .assign() for chainable column additions and you stop reassigning to df at every step. Method chaining makes notebooks read like a recipe.

When to leave pandas

Above ten million rows, pandas becomes painful. Reach for DuckDB (SQL on Parquet, brilliant), Polars (pandas-like API, much faster), or just push the work into the warehouse. Knowing when to step off pandas is itself a senior signal.

Practice tip

Take a Kaggle dataset and redo any familiar tutorial entirely with method chains. The discipline forces these one-liners into muscle memory.