Sorting · Pandas · CodeMIRAI

sort_values

Python

import pandas as pd

df = pd.DataFrame({
    "name": ["Asha", "Leo", "Mei", "Raj", "Zoe", "Ana"],
    "course": ["Python", "SQL", "Python", "Java", "SQL", "Java"],
    "score": [92, 78, 85, 88, 95, 81],
})
print(df.sort_values("score"))                      # lowest first
print(df.sort_values("score", ascending=False))     # highest first

Sorting returns a new DataFrame; df itself is unchanged unless you assign the result back: df = df.sort_values("score").

Sorting by several columns

Python

import pandas as pd

df = pd.DataFrame({
    "name": ["Asha", "Leo", "Mei", "Raj", "Zoe", "Ana"],
    "course": ["Python", "SQL", "Python", "Java", "SQL", "Java"],
    "score": [92, 78, 85, 88, 95, 81],
})
print(df.sort_values(["course", "score"], ascending=[True, False]))

Courses alphabetically, and within each course, the highest score first. ascending takes one value per column.

Text and missing values

Python

import numpy as np
import pandas as pd

df = pd.DataFrame({"name": ["mei", "Asha", "leo", "Zoe"], "score": [85, np.nan, 78, 95]})
print(df.sort_values("name"))                                  # capitals sort before lower case
print(df.sort_values("name", key=lambda s: s.str.lower()))     # case-insensitive
print(df.sort_values("score", na_position="first"))            # missing values first

Plain text sorting puts every capital letter before every lower-case one. A key function transforms the values just for the comparison. Missing values go last unless you say otherwise.

Sorting by the index

Python

import pandas as pd

scores = pd.DataFrame({"score": [92, 78, 85]}, index=["Mei", "Asha", "Leo"])
print(scores.sort_index())                     # alphabetical by label
print(scores.sort_index(ascending=False))

Top and bottom rows

Python

import pandas as pd

df = pd.DataFrame({
    "name": ["Asha", "Leo", "Mei", "Raj", "Zoe", "Ana"],
    "score": [92, 78, 85, 88, 95, 81],
})
print(df.nlargest(3, "score"))       # the top three
print(df.nsmallest(2, "score"))      # the bottom two

nlargest and nsmallest read more clearly than sorting and taking head, and are faster on big tables.

Ranking

Python

import pandas as pd

scores = pd.Series([95, 88, 88, 72], index=["Zoe", "Raj", "Ana", "Leo"])
print(pd.DataFrame({
    "score": scores,
    "min": scores.rank(ascending=False, method="min"),
    "dense": scores.rank(ascending=False, method="dense"),
    "average": scores.rank(ascending=False),
}))

A rank says where each value stands. Ties need a rule: min gives tied values the best rank they share (1, 2, 2, 4, like sports tables), dense leaves no gaps (1, 2, 2, 3), and the default, average, gives each the mean of their positions (2.5).

Renumbering: reset_index

Python

import pandas as pd

df = pd.DataFrame({"name": ["Asha", "Leo", "Mei"], "score": [92, 78, 85]})
ranked = df.sort_values("score", ascending=False)
print(ranked)                              # the index still has the old row numbers
print(ranked.reset_index(drop=True))       # renumbered from 0

After sorting or filtering, the index keeps the original labels. reset_index(drop=True) renumbers the rows from 0 and throws the old index away.