How to find missing values in pandas

Finding missing values in pandas before cleanup shows which records need review and which columns will change if data is dropped, filled, grouped, or exported. DataFrame.isna() builds a boolean mask that marks missing cells, so the same check can drive column counts, row filters, and required-field checks.

pandas treats None, numpy.nan, pd.NaT, and nullable pd.NA as missing values. Empty strings and infinity remain ordinary values unless an import or preprocessing step converts them to missing markers.

A DataFrame named df is enough for column and row inspection. Use notna() when the next operation should keep only present values, and use isna() when the next operation should inspect or isolate gaps.

1. Start the Python 3 interpreter in an environment with pandas installed.

   $ python3
   Python 3.13.14
   >>>

2. Import pandas and NumPy.

   >>> import pandas as pd
   >>> import numpy as np

3. Create or load the DataFrame that needs checking.

   >>> df = pd.DataFrame(
   ...     {
   ...         "customer": ["Ava", "Ben", "Cy", "Dana"],
   ...         "email": ["ava@example.net", None, "", "dana@example.net"],
   ...         "signup": [
   ...             pd.Timestamp("2026-01-05"),
   ...             pd.NaT,
   ...             pd.Timestamp("2026-01-07"),
   ...             pd.Timestamp("2026-01-08"),
   ...         ],
   ...         "amount": pd.Series([120.0, np.nan, 75.5, 0.0], dtype="float64"),
   ...         "plan": pd.Series(["pro", "basic", pd.NA, "pro"], dtype="string"),
   ...     }
   ... )
   >>> df
     customer             email     signup  amount   plan
   0      Ava   ava@example.net 2026-01-05   120.0    pro
   1      Ben               NaN        NaT     NaN  basic
   2       Cy                   2026-01-07    75.5   <NA>
   3     Dana  dana@example.net 2026-01-08     0.0    pro

   Use the same variable name for an imported CSV, Excel, SQL, or Parquet DataFrame when checking real data.

4. Show the missing-value mask for each cell.

   >>> df.isna()
      customer  email  signup  amount   plan
   0     False  False   False   False  False
   1     False   True    True    True  False
   2     False  False   False   False   True
   3     False  False   False   False  False

   Do not use equality checks against np.nan or pd.NA to find missing cells. Missing sentinels do not compare like ordinary values.

5. Count missing values in each column.

   >>> df.isna().sum()
   customer    0
   email       1
   signup      1
   amount      1
   plan        1
   dtype: int64

6. List rows that contain at least one missing value.

   >>> df[df.isna().any(axis="columns")]
     customer email     signup  amount   plan
   1      Ben   NaN        NaT     NaN  basic
   2       Cy       2026-01-07    75.5   <NA>

   axis=“columns” checks across each row. The default axis checks down each column.

7. Set the required columns for a complete-row check.

   >>> required = ["customer", "amount"]

8. Keep rows where required columns are present.

   >>> df[df[required].notna().all(axis="columns")]
     customer             email     signup  amount  plan
   0      Ava   ava@example.net 2026-01-05   120.0   pro
   2       Cy                   2026-01-07    75.5  <NA>
   3     Dana  dana@example.net 2026-01-08     0.0   pro

   Change the column list to the fields that must be present before analysis, export, or reporting.

9. Build a small series of candidate missing-value markers.

   >>> sentinels = pd.Series(
   ...     [None, np.nan, pd.NA, pd.NaT, "", np.inf],
   ...     index=["None", "np.nan", "pd.NA", "pd.NaT", "empty string", "np.inf"],
   ...     dtype="object",
   ... )

10. Check which marker values pandas treats as missing.

    >>> pd.DataFrame({"value": sentinels, "is_missing": pd.isna(sentinels)})
                 value  is_missing
    None          None        True
    np.nan         NaN        True
    pd.NA         <NA>        True
    pd.NaT         NaT        True
    empty string             False
    np.inf         inf       False

    Empty strings and np.inf return False here. Normalize them before using isna() if a source system uses those values to mean missing data.