Fancy indexing selects several positions or labels at once by placing an index collection inside square brackets. With a Series, s.loc[[2, 4]] returns both indexed elements while preserving their original labels. With a DataFrame, df.loc uses the first argument for rows and the second for columns, so lists can select either dimension. section 1.13 section 2.8
Use fancy indexing with pandas objects to:
- Retrieve rows a and c from column x with df.loc[['a', 'c'], 'x']. section 2.8
- Retrieve row a from columns v and y, or combine lists to select multiple rows and columns. section 2.8
- Combine label lists with boolean indexes, such as selecting rows where x exceeds 200 and columns where c exceeds 135. section 2.8
- Distinguish a list of labels from a mask index: labels name positions, while True and False values select matching positions. section 1.13
Fill missing sales for several products
After adding products at indexes 5, 6, and 7, select their sales column with df.loc[[5, 6, 7], 'sales']. Assigning [100, 200, 75] updates all three missing values in one operation. A scalar on the right would instead be broadcast to every matching cell. section 2.10 section 2.8
df.loc[['a', 'c'], #1
['v','y']] #2
Inspect the two-list loc expression that selects rows a and c together with columns v and y. section 2.8
