Rename, Drop & Reorder

Raw data rarely comes with tidy column names or only the columns you need. A CSV may call a column " Player Name" (with a leading space), include ten columns you never use, and put the important ones last. This topic covers the three clean-up tools you use on almost every dataset: rename, drop, and reorder.

Sample table:

           player team  runs  balls  sixes
0    Shubman Gill   GT   890    564     33
1  Faf du Plessis  RCB   730    475     36
2     Virat Kohli  RCB   639    457     16
3     Rinku Singh  KKR   474    317     29

Definition

rename changes column names and/or index labels using a mapping (dict) or a function. drop removes rows or columns by label. Reordering means selecting the columns in a new order. All three return a new DataFrame and leave the original unchanged (unless you reassign it).

• Mapping — a dict {"old": "new"}; only the names you mention change.
• By label — drop works with names/labels, not positions (use df.columns[i] or df.index[i] to translate a position).
• Return a new DataFrame — write df = df.drop(...). (inplace=True exists but is discouraged.)

Syntax

df.rename(columns={"old": "new"}, index={old_label: new_label}, errors="ignore")
df.columns = [list of all new names]          # replace every name at once

df.drop(columns=[...])                        # drop columns
df.drop(index=[...])                          # drop rows by label
df.drop(labels, axis=0 | 1, errors="raise")   # older style
del df["col"]                                 # delete a column in place
df.pop("col")                                 # remove a column in place AND return it

df[["c3", "c1", "c2"]]                        # reorder by selecting
df.reindex(columns=[...])                     # reorder; unknown names -> NaN column

Example — rename columns and index labels

r = df.rename(columns={"player": "batter", "runs": "total_runs"})
print(r)
print(df.columns.tolist())
print(df.rename(index={0: "first", 3: "last"}))

           batter team  total_runs  balls  sixes
0    Shubman Gill   GT         890    564     33
1  Faf du Plessis  RCB         730    475     36
2     Virat Kohli  RCB         639    457     16
3     Rinku Singh  KKR         474    317     29
['player', 'team', 'runs', 'balls', 'sixes']
               player team  runs  balls  sixes
first    Shubman Gill   GT   890    564     33
1      Faf du Plessis  RCB   730    475     36
2         Virat Kohli  RCB   639    457     16
last      Rinku Singh  KKR   474    317     29

How it works: only the keys you mention are renamed. The original df still has its old names — rename returned a new DataFrame (r). Passing a function renames every name: df.rename(columns=str.upper) → ['PLAYER', 'TEAM', 'RUNS', 'BALLS', 'SIXES'].

Example — clean messy column names in one line

messy = pd.DataFrame(columns=[" Player Name", "Team ", "Total Runs"])
messy.columns = messy.columns.str.strip().str.lower().str.replace(" ", "_")
print(messy.columns.tolist())

['player_name', 'team', 'total_runs']

How it works: df.columns is an Index of strings, so it supports the .str methods: strip removes outer spaces, lower lowercases, replace swaps spaces for underscores. Assigning to df.columns replaces all names — the list must have exactly as many names as there are columns, otherwise ValueError: Length mismatch: Expected axis has 5 elements, new values have 2 elements.

Example — drop columns and rows

print(df.drop(columns=["balls", "sixes"]))
print(df.drop(index=[1, 3]))

           player team  runs
0    Shubman Gill   GT   890
1  Faf du Plessis  RCB   730
2     Virat Kohli  RCB   639
3     Rinku Singh  KKR   474
         player team  runs  balls  sixes
0  Shubman Gill   GT   890    564     33
2   Virat Kohli  RCB   639    457     16

How it works: columns= and index= make it obvious what you are dropping. The older form df.drop(["balls"], axis=1) does the same as df.drop(columns=["balls"]), and df.drop([0]) drops the row labelled 0 (axis 0 is the default). Note the remaining rows keep their original labels (0, 2) — they are not renumbered.

Dropping a name that doesn't exist raises KeyError: "['wickets'] not found in axis"; add errors="ignore" to skip it silently.

Example — drop rows by a condition

print(df.drop(df[df["team"] == "RCB"].index))

         player team  runs  balls  sixes
0  Shubman Gill   GT   890    564     33
3   Rinku Singh  KKR   474    317     29

df[df["team"] == "RCB"].index gives the labels of RCB rows, and drop removes them. The more common way to write the same thing is to keep the rows you want: df[df["team"] != "RCB"].

Example — del and pop

d = df.copy()
del d["sixes"]
balls = d.pop("balls")
print(balls.tolist(), d.columns.tolist())

[564, 475, 457, 317] ['player', 'team', 'runs']

del and pop change the DataFrame in place. pop also returns the removed column as a Series — handy for separating a target column (y = df.pop("price")) in machine learning.

Example — reorder columns

print(df[["team", "player", "sixes", "runs", "balls"]])
cols = ["runs"] + [c for c in df.columns if c != "runs"]
print(df[cols].columns.tolist())

  team          player  sixes  runs  balls
0   GT    Shubman Gill     33   890    564
1  RCB  Faf du Plessis     36   730    475
2  RCB     Virat Kohli     16   639    457
3  KKR     Rinku Singh     29   474    317
['runs', 'player', 'team', 'balls', 'sixes']

How it works: selecting a list of columns returns them in the list's order. To move one column to the front without typing all the others, build the list with a comprehension. Other options: df.reindex(columns=[...]) (unknown names become all-NaN columns instead of raising), df.sort_index(axis=1) (alphabetical), and df.loc[:, "team":"balls"] (a label range of columns).

Comparison

Common mistakes

• ❌ df.rename(columns={"runs": "total"}) and then using df["total"] → KeyError. ✅ df = df.rename(...) — rename returns a new DataFrame.
• ❌ df.drop("balls") → KeyError (looks for a row called "balls"). ✅ df.drop(columns="balls").
• ❌ df.rename({"runs": "total"}) without columns=. ✅ A bare dict renames the index by default; always say columns=.
• ❌ df = df.drop(columns="x", inplace=True) → df becomes None. ✅ Choose one: df = df.drop(columns="x").
• ❌ Expecting row labels to be renumbered after a drop. ✅ Use df.reset_index(drop=True) (covered in the indexing chapter).

Interview tip

Q: How do you drop a column in pandas, and what's the difference between drop, del and pop? A: df.drop(columns=["c"]) returns a new DataFrame without the column (the original is unchanged unless reassigned). del df["c"] removes it in place and returns nothing. df.pop("c") removes it in place and returns the removed column as a Series.