Convert text to numbers
A column contains numeric text that must participate in arithmetic or numeric filtering.
Polars · Module
.cast
Expr.cast converts an expression to a specified data type. Failed casts raise by default, or become null with strict=False; numeric overflow can be wrapped explicitly.
How it works
Expr.cast changes the data type of an expression while keeping the operation inside Polars' expression pipeline. By default, invalid conversions raise an exception; setting strict=False turns invalid results into nulls, while wrap_numerical=True permits overflowing numeric casts to wrap.
Syntax
Expr.cast(dtype: PolarsDataType | DataTypeExpr | type[Any], *, strict: bool = True, wrap_numerical: bool = False) → ExprParameters
| Name | Type | Description |
|---|---|---|
dtype | PolarsDataType | DataTypeExpr | type[Any] | Data type to cast to. |
strict | bool = True | Raise if cast is invalid; when false, invalid casts produce null values. |
wrap_numerical | bool = False | Wrap overflowing numeric casts instead of marking them as invalid. |
Returns
Expr
Key points
- The method converts an expression from its current data type to a requested target type.
- The target type can be supplied as a Polars data type, a data-type expression, or a Python type.
- Casting is strict by default, so conversion failures are not silently ignored.
- Numeric overflow wrapping is opt-in and disabled by default.
Common mistakes
- With
strict=False, values that cannot be converted do not raise an exception; Polars produces null values instead. - For numeric conversions,
wrap_numerical=Truechanges overflow handling from invalid-cast behavior to wrapping.
Example: Casting expression columns
import polars as pl
df = pl.DataFrame({"a": [1, 2, 3], "b": ["4", "5", "6"]})
result = df.with_columns(
pl.col("a").cast(pl.Float64),
pl.col("b").cast(pl.Int32),
)
print(result)Casts two expression columns to different target data types.
When to use it
Standardize numeric types
A pipeline needs consistent numeric precision or integer width across expressions.
Handle invalid input safely
Input data may contain malformed values that should become missing values instead of stopping processing.
Interview questions
QWhat does Expr.cast do?
Expr.cast converts an expression to a requested data type and returns another expression. The target may be a Polars data type, a data-type expression, or a Python type.
Follow-up: How would you preserve a lazy expression pipeline while converting a string column? Use pl.col("name").cast(pl.Int32) inside select or with_columns.
QHow does strictness affect failed casts?
The default is strict=True, so an invalid conversion raises an exception. With strict=False, invalid casts produce null values instead.
Follow-up: When is non-strict casting useful? It is useful when malformed input should be retained in the result as missing data rather than aborting the operation.
QWhat is the purpose of wrap_numerical?
wrap_numerical controls overflow behavior for numeric casts. When enabled, overflowing numeric values wrap instead of being marked as invalid.
Follow-up: What should be considered before enabling it? Verify that wraparound semantics are acceptable, because the resulting value may differ substantially from the original.
Related
polars.Series.castpolars.DataFrame.castpolars.colpolars.Expr