gluonts.transform.sampler module

class gluonts.transform.sampler.BucketInstanceSampler(*, axis: int = - 1, min_past: int = 0, min_future: int = 0, scale_histogram: gluonts.dataset.stat.ScaleHistogram)

Bases: gluonts.transform.sampler.InstanceSampler

This sample can be used when working with a set of time series that have a skewed distributions. For instance, if the dataset contains many time series with small values and few with large values.

The probability of sampling from bucket i is the inverse of its number of elements.

Parameters

scale_histogram (gluonts.dataset.stat.ScaleHistogram) – The histogram of scale for the time series. Here scale is the mean abs value of the time series.

axis: int

min_future: int

min_past: int

scale_histogram: gluonts.dataset.stat.ScaleHistogram

class gluonts.transform.sampler.ContinuousTimePointSampler(*, min_past: float = 0.0, min_future: float = 0.0)

Bases: pydantic.main.BaseModel

Abstract class for “continuous time” samplers, which, given a lower bound and upper bound, sample “points” (events) in continuous time from a specified interval.

min_future: float

min_past: float

class gluonts.transform.sampler.ContinuousTimePredictionSampler(*, min_past: float = 0.0, min_future: float = 0.0, allow_empty_interval: bool = False)

Bases: gluonts.transform.sampler.ContinuousTimePointSampler

allow_empty_interval: bool

min_future: float

min_past: float

class gluonts.transform.sampler.ContinuousTimeUniformSampler(*, min_past: float = 0.0, min_future: float = 0.0, num_instances: int)

Bases: gluonts.transform.sampler.ContinuousTimePointSampler

Implements a simple random sampler to sample points in the continuous interval between a and b.

min_future: float

min_past: float

num_instances: int

class gluonts.transform.sampler.ExpectedNumInstanceSampler(*, axis: int = - 1, min_past: int = 0, min_future: int = 0, num_instances: float, total_length: int = 0, n: int = 0)

Bases: gluonts.transform.sampler.InstanceSampler

Keeps track of the average time series length and adjusts the probability per time point such that on average num_instances training examples are generated per time series.

Parameters

num_instances (float) – number of training examples generated per time series on average

axis: int

min_future: int

min_past: int

n: int

num_instances: float

total_length: int

class gluonts.transform.sampler.InstanceSampler(*, axis: int = - 1, min_past: int = 0, min_future: int = 0)

Bases: pydantic.main.BaseModel

An InstanceSampler is called with the time series ts, and returns a set of indices at which training instances will be generated.

The sampled indices i satisfy a <= i <= b, where a = min_past and b = ts.shape[axis] - min_future.

class Config

Bases: object

arbitrary_types_allowed = True

axis: int

min_future: int

min_past: int

class gluonts.transform.sampler.PredictionSplitSampler(*, axis: int = - 1, min_past: int = 0, min_future: int = 0, allow_empty_interval: bool = False)

Bases: gluonts.transform.sampler.InstanceSampler

Sampler used for prediction.

Always selects the last time point for splitting i.e. the forecast point for the time series.

allow_empty_interval: bool

gluonts.transform.sampler.TestSplitSampler(axis: int = - 1, min_past: int = 0) → gluonts.transform.sampler.PredictionSplitSampler

class gluonts.transform.sampler.UniformSplitSampler(*, axis: int = - 1, min_past: int = 0, min_future: int = 0, p: float)

Bases: gluonts.transform.sampler.InstanceSampler

Samples each point with the same fixed probability.

Parameters

p (float) – Probability of selecting a time point

axis: int

min_future: int

min_past: int

p: float

gluonts.transform.sampler.ValidationSplitSampler(axis: int = - 1, min_past: int = 0, min_future: int = 0) → gluonts.transform.sampler.PredictionSplitSampler