Pandas flashcards, generated for you.

A DataFrame is a 2D table with rows and columns (like SQL table or Excel sheet). A Series is 1D labeled array (single column). DataFrames are made of multiple Series.

Use df['column_name'] or df.column_name. Returns a Series. Use df[['column_name']] with double brackets to keep it as a DataFrame.

.loc uses label-based indexing (row/column names). .iloc uses integer position-based indexing (0, 1, 2, ...). Example: df.loc['row1', 'col1'] vs df.iloc[0, 0]

Use boolean indexing: df[df['column'] > value]. Returns DataFrame with only rows where condition is True. Chain conditions with & (and), | (or), ~ (not).

.groupby(column) splits data by unique values in a column, creating groups. Chain with aggregation: df.groupby('category').sum(), .mean(), .count(), etc. Returns grouped results.

pd.merge(df1, df2, on='key') joins on shared column. join types: 'inner' (matching keys only), 'outer' (all keys), 'left' (all from df1), 'right' (all from df2).

.merge() joins on column values (more flexible). .join() joins on indices by default. .join() is shorthand: df1.join(df2) merges by index. Both can use 'on' or 'left_on'/'right_on'.

.isna() or .isnull() detects NaN. .dropna() removes rows/columns with NaN. .fillna(value) replaces NaN. .fillna(method='ffill') forward-fills, .fillna(method='bfill') back-fills.

.apply(func) applies a function to each row/column. Use for complex logic, custom functions, or operations pandas doesn't optimize. Slower than vectorized ops but more flexible. df.apply(lambda x: x + 1)

.pivot(index, columns, values) reshapes data (no aggregation). Errors if duplicates exist. .pivot_table(index, columns, values, aggfunc='mean') reshapes and aggregates duplicates. pivot_table is safer and more common.