pandas Cheat Sheet

This page contains a condensed overview of pandas, the go-to Python library for working with tabular data. It covers loading and inspecting data, selecting and filtering rows and columns, cleaning messy values, grouping and aggregating, and plotting your results. You can also download the information as a printable cheat sheet:

Free Bonus: pandas Cheat Sheet

Get a pandas Cheat Sheet (PDF) and keep the essentials for loading, inspecting, selecting, cleaning, grouping, and plotting your data at hand:

pandas Cheat Sheet

Practice with hands-on coding exercises, quizzes, and guided learning paths. Not sure where to begin? Start here.

New to pandas?

Loading Data

  • Import pandas as pd; everyone does
  • A DataFrame is a table of Series columns that share one index
  • read_csv() also takes URLs and zipped files
  • pandas 3 needs Python 3.11 or newer
Shell Install pandas 
$ python -m pip install pandas matplotlib
Python Read a CSV File 
import pandas as pd

df = pd.read_csv("sales.csv")
df = pd.read_csv(
    "sales.csv",
    usecols=["date", "region", "units"],
    parse_dates=["date"],
)
Python Build a DataFrame by Hand 
df = pd.DataFrame({
    "city": ["Oslo", "Lima"],
    "temp": [4, 19],
})
Python Read Other Formats 
pd.read_excel("sales.xlsx")  # Needs openpyxl
pd.read_json("sales.json")
pd.read_parquet("sales.parquet")
Python Save Your Results 
df.to_csv("clean.csv", index=False)

Want to go deeper on loading data?

Inspecting Data

  • A non-null count below the row count means missing values
  • Integer columns with gaps load as float64
Python Take a First Look 
df.head(3)     # First 3 rows
df.tail()      # Last 5 rows
df.sample(2)   # 2 random rows
df.shape       # (8, 5)
df.columns     # Column labels
Python Check Types and Stats 
df.info()      # Dtypes + non-null counts
df.dtypes      # Units → float64
df.describe()  # Count, mean, std, ...
Python Count Categories 
>>> df["region"].value_counts()
region
North    3
South    3
East     2
Name: count, dtype: int64

Not sure what to look for first?

Selecting Data

  • .loc uses labels, and its slices include the end
  • .iloc uses positions, and its slices exclude the end
  • One label gives a Series, a list gives a DataFrame
Python Pick Columns 
df["units"]              # Series
df[["region", "units"]]  # DataFrame
Python Pick Rows and Cells 
df.loc[0, "region"]    # 'North'
df.loc[0:2, "units"]   # Rows 0, 1, and 2
df.iloc[0, 1]          # 'North'
df.iloc[:2, -1]        # First 2 prices
df.iloc[-1]            # Last row

Counted the rows?

Filtering and Sorting

  • Combine conditions with &, |, and ~, and wrap each one in parentheses
  • Use @name to reach Python variables in .query()
Python Filter With Boolean Masks 
df[df["units"] > 10]           # 3 rows
df[(df["region"] == "North")
   & (df["units"] > 5)]        # 2 rows
df[df["product"].isin(["Widget", "Gizmo"])]
df[df["product"].str.startswith("G")]
Python Filter With a Query String 
min_units = 8
df.query("units >= @min_units")
df.query("region == 'East' and units > 10")
Python Sort Rows 
df.sort_values("units", ascending=False)
df.nlargest(3, "units")      # Top 3 rows

Think you can filter any DataFrame?

Cleaning Data

  • Most methods return a new object, so assign the result back
  • Change cells with .loc; pandas 3 ignores chained assignment like df["a"][0] = 1
  • pd.to_numeric(s, errors="coerce") turns bad values into NaN
Python Find Missing Values 
>>> df.isna().sum()
date       0
region     0
product    0
units      1
price      1
dtype: int64
Python Drop or Fill Missing Values 
df.dropna()                  # 6 rows left
df.dropna(subset=["units"])  # 7 rows left
df["units"] = df["units"].fillna(0)
df["price"] = df["price"].fillna(
    df["price"].median()
)
Python Fix Data Types 
df["date"] = pd.to_datetime(df["date"])
df["units"] = df["units"].astype(int)
df["region"] = (
    df["region"].astype("category")
)
Python Tidy Up and Add Columns 
df = df.drop_duplicates()
df["region"] = df["region"].str.strip()
df = df.rename(columns={"qty": "units"})
df["revenue"] = df["units"] * df["price"]
df.loc[0, "units"] = 13  # Set one cell
Python Unify Inconsistent Labels 
df["region"] = df["region"].replace(
    {"N": "North", "S": "South"}
)

Wondering why your change didn’t stick?

Grouping and Aggregating

  • .groupby() splits rows, applies a function, and combines the results
  • Add .reset_index() to get a flat table back
  • Month-end resampling uses "ME" since pandas 2.2
Python Aggregate One Column 
>>> df.groupby("region")["units"].sum()
region
East     23
North    24
South    24
Name: units, dtype: int64
Python Compute Several Stats 
df.groupby("region").agg(
    total=("revenue", "sum"),
    orders=("units", "count"),
)
Python Build a Pivot Table 
df.pivot_table(
    index="region", columns="product",
    values="units", aggfunc="sum",
)
Python Group by Time 
df.resample("D", on="date")["units"].sum()

Still fuzzy on groupby?

Plotting

  • .plot() draws with Matplotlib, so install it too
  • Notebooks show plots inline; scripts need plt.show()
Python From Raw CSV to Chart 
import matplotlib.pyplot as plt

# Which product earns the most?
(pd.read_csv("sales.csv")
   .dropna(subset=["units", "price"])
   .assign(revenue=lambda d:
           d["units"] * d["price"])
   .groupby("product")["revenue"]
   .sum()
   .sort_values()
   .plot(kind="barh", title="Revenue"))
plt.tight_layout()
plt.savefig("revenue.png")
Python Pick a Plot Kind 
df.plot(x="date", y="units")  # Line
df["units"].plot(kind="hist")
df.plot.scatter(x="price", y="units")
df["product"].value_counts().plot.pie()

Want charts that tell a better story?

Ready to go beyond the cheat sheet?

You can download this information as a printable cheat sheet:

Free Bonus: pandas Cheat Sheet

Get a pandas Cheat Sheet (PDF) and keep the essentials for loading, inspecting, selecting, cleaning, grouping, and plotting your data at hand:

pandas Cheat Sheet