Skip to content

Series API

Bodo DataFrames supports Pandas Series methods and accessors that are listed below. They can be accessed through BodoSeries and follow the same behavior as their Pandas equivalents. For details on usage, we link to the corresponding Pandas documentation.

Note

If the user code encounters an unsupported Pandas API or an unsupported parameter, Bodo Bodo DataFrames gracefully falls back to native Pandas. See overview of Bodo DataFrames for more info.

AI Functions

Binary operator functions

Computations / descriptive stats

Note

Unlike pandas’, the quantile in Bodo DataFrames is an approximate quantile based on the KLL sketch algorithm with probabilistic error guarantees, since computing exact quantiles across large datasets is extremely expensive.

Note

For the following reduction methods, only default parameters are currently supported.

Datetimelike properties

Note

Input must be a Series of datetime-like data.

Datetime properties

Note

For missing datetime values (NaT), Bodo's datetime predicate accessors (e.g., .is_month_end, .is_leap_year) return <NA> to preserve nullability, whereas Pandas returns False.

Datetime methods

Note

Locale format must be strict: The locale parameter in month_name and day_name must follow the exact system locale naming convention (e.g., "pt_BR.UTF-8" or "en_US.utf-8"). Variants like "pt_BR.utf8" may not be recognized and trigger an error.

Note

Bodo currently only supports "NaT" for the ambiguous parameter in tz_localize. "raise", "infer", or boolean arrays are not supported and will trigger a fallback to Pandas. Similarly, for the nonexistent parameter, "raise" is not supported and will trigger a fallback. Due to these limitations, the default behavior in Bodo is ambiguous="NaT" and nonexistent="NaT".

Function application

Missing data handling

Reindexing / Selection / Label manipulation

Reshaping, sorting

String handling

Note

cat falls back to Pandas when the others parameter is not specified (i.e., others=None).