Skip to content

API Reference#

This page gives an overview of all of Temporian's public symbols.

Check the index on the left for a more detailed description of any symbol.


Symbol Description
tp.EventSetNode Reference to the input or output of an operator in the compute graph.
tp.EventSet Container for actual temporal data.
tp.Schema Description of the data inside an EventSetNode or EventSet.
tp.FeatureSchema Description of a feature inside a Schema.
tp.IndexSchema Description of an index inside a Schema.


Symbol Description Evaluates EventSetNodes on EventSets.
tp.plot() Plots EventSets.
tp.event_set() Creates an EventSet from arrays (lists, NumPy arrays, Pandas Series.)
tp.input_node() Creates an input EventSetNode, that can be used to feed data into a graph.


Symbol Description
tp.from_pandas() Converts a Pandas DataFrame into an EventSet.
tp.to_pandas() Converts an EventSet to a pandas DataFrame.
tp.from_csv() Reads an EventSet from a CSV file.
tp.to_csv() Saves an EventSet to a CSV file.


Symbols Description
tp.duration.milliseconds() tp.duration.seconds() tp.duration.minutes() tp.duration.hours() tp.duration.days() tp.duration.weeks() Convert input value from milliseconds / seconds / minutes / hours / days / weeks to a Duration in seconds.


Symbols Description
tp.combine() Combines events from EventSets with different samplings.
tp.glue() Concatenates features from EventSets with the same sampling.
EventSet.abs() Computes the absolute value of the features.
EventSet.add_index() Adds indexes to an EventSet.
EventSet.arccos() Computes the inverse cosine of the features.
EventSet.arcsin() Computes the inverse sine of the features.
EventSet.arctan() Computes the inverse tangent of the features.
EventSet.begin() Generates a single timestamp at the beginning of the input.
EventSet.cast() Casts the dtype of features.
EventSet.cos() Computes the cosine of the features.
EventSet.drop_index() Removes indexes from an EventSet.
EventSet.end() Generates a single timestamp at the end of the input.
EventSet.enumerate() Creates an ordinal feature enumerating the events according to their timestamp.
EventSet.equal() Creates boolean features with event-wise equality to another EventSet or to a scalar value.
EventSet.experimental_fast_fourier_transform() Applies a Fast Fourier Transform.
EventSet.filter() Filters out events in an EventSet for which a condition is false.
EventSet.filter_empty_index() Filters out indexes without events.
EventSet.filter_moving_count() Skips events such that no more than one event is within a time window of window_length.
EventSet.isnan() Event-wise boolean that is True in the NaN positions of the input events. Equivalent to ~evset.notnan().
EventSet.join() Joins EventSets with different samplings but the same index together.
EventSet.lag() Adds a delay to an EventSet's timestamps.
EventSet.leak() Subtracts a duration from an EventSet's timestamps.
EventSet.log() Computes natural logarithm of the features Applies a function on each of an EventSet's values.
EventSet.notnan() Event-wise boolean that is True if the input values are not NaN. Equivalent to ~evset.isnan().
EventSet.prefix() Adds a prefix to the names of the features in an EventSet.
EventSet.propagate() Propagates feature values over a sub index.
EventSet.rename() Renames an EventSet's features and index.
EventSet.resample() Resamples an EventSet at each timestamp of another EventSet. Selects a subset of features from an EventSet.
EventSet.select_index_values() Selects a subset of index values from an EventSet.
EventSet.set_index() Replaces the indexes in an EventSet.
EventSet.sin() Computes the sine of the features.
EventSet.since_last() Computes the amount of time since the last distinct timestamp.
EventSet.tan() Computes the tangent of the features.
EventSet.tick() Generates timestamps at regular intervals in the range of a guide.
EventSet.tick_calendar() Generates timestamps at the specified calendar date-time events.
EventSet.timestamps() Creates a feature from the events timestamps (float64).
EventSet.unique_timestamps() Removes events with duplicated timestamps from an EventSet.
EventSet.until_next() Duration until the next sampling event.
EventSet.where() Choose events from two possible sources, based on boolean conditions.

Calendar operators#

Symbols Description
EventSet.calendar_day_of_month() EventSet.calendar_day_of_week() EventSet.calendar_day_of_year() EventSet.calendar_hour() EventSet.calendar_iso_week() EventSet.calendar_minute() EventSet.calendar_month() EventSet.calendar_second() EventSet.calendar_year() Obtain the day of month / day of week / day of year / hour / ISO week / minute / month / second / year the timestamps in an EventSet are in.

Window operators#

Symbols Description
EventSet.simple_moving_average() EventSet.moving_standard_deviation() EventSet.cumsum() EventSet.moving_sum() EventSet.moving_count() EventSet.moving_min() EventSet.moving_max() EventSet.cumprod() EventSet.moving_product() Compute an operation on the values in a sliding window over an EventSet's timestamps.

Python operators#

Symbols Description
+ (add), - (subtract), \* (multiply), / (divide), // (floor divide), % (modulo), ** (power) Event-wise arithmetic operations between two EventSets or with a scalar number. See the corresponding User Guide section for more info.
!= (not equal), > (greater than), >= (greater or equal), < (less), <= (less or equal) Event-wise comparison between EventSets or to a scalar number. Note that == is not supported, use EventSet.equal() instead. See the User Guide for more info.
& (and), \| (or), ^ (xor) Event-wise logic operators between boolean EventSets.