DataSet

class minkit.DataSet(data, pars, weights=None)

Bases: minkit.DataObject

Definition of an unbinned data set to evaluate PDFs.

Parameters:

• data (darray) – data in the provided backend.
• pars (Registry(Parameter)) – data parameters.
• weights (darray or None) – possible set of weights.

Attributes Summary

aop	Object to do operations on arrays.
backend	Backend interface.
data_pars	Data parameters associated to this sample.
ndim	Number of dimensions.
values	Values of the data set.
weights	Weights of the sample.

Methods Summary

from_ndarray(arr, arg[, weights, backend])	Build the class from a single array.
from_records(arr, data_pars[, weights, backend])	Build the class from a numpy.ndarray object.
get(index)	Get the values given an index.
make_binned([bins])	Make a binned version of this sample.
merge(samples[, maximum])	Merge many samples into one.
subset(arg[, rescale_weights])	Get a subset of this data set.
to_backend(backend)	Initialize this class in a different backend.
to_records()	Convert this class into a numpy.ndarray object.

Attributes Documentation

aop

Object to do operations on arrays.

Type:ArrayOperations

backend

Backend interface.

Type:Backend

data_pars

Data parameters associated to this sample.

Type:Registry(Parameter)

ndim

Number of dimensions.

Type:int

values

Values of the data set.

Type:darray

weights

Weights of the sample.

Type:darray or None

Methods Documentation

classmethod from_ndarray(arr, arg, weights=None, backend=None)

Build the class from a single array.

Parameters:

• arr (numpy.ndarray) – array of data.
• arg (Registry(Parameter)) – if arr only contains one set of values, it must be a single data parameter. Otherwise a collection of parameters.
• weights (numpy.ndarray or None) – possible weights to use.
• backend (Backend or None) – backend where the data set is built.

classmethod from_records(arr, data_pars, weights=None, backend=None)

Build the class from a numpy.ndarray object.

Parameters:

• arr (numpy.ndarray) – array of data.
• data_pars (Registry(Parameter)) – data parameters.
• weights (numpy.ndarray or None) – possible weights to use.
• backend (Backend or None) – backend where the data set is built.

Raises:

RuntimeError – If a parameter is not found in the input array.

get(index)

Get the values given an index.

Parameters:index (int) – index to process. Returns:Values at the index. Return type:numpy.ndarray

make_binned(bins=100)

Make a binned version of this sample.

Parameters:bins (int or tuple(int, ..)) – number of bins per data parameter. Returns:Binned data sample. Return type:BinnedDataSet

classmethod merge(samples, maximum=None)

Merge many samples into one. If maximum is specified, then the last elements will be dropped.

Parameters:

• samples (tuple(DataSet)) – samples to merge.
• maximum (int or None) – maximum number of entries for the final sample.

Returns:

Merged sample.

Return type:

DataSet

Raises:

RuntimeError – If the samples have different parameters or if only some of them have weights.

… warning:: If maximum is specified, the last elements corresponding to the

last samples might be dropped.

subset(arg, rescale_weights=False)

Get a subset of this data set. If arg is a string, it will be considered as a range. In case it is a barray, then it is considered to be a mask array. If rescale_weights is set to True, then the weights are rescaled so their statistical weight in minimization processes is proportional to the event weights:

\[\omega^\prime_i = \omega_i \times \frac{\sum_{j = 0}^n \omega_j}{\sum_{j = 0}^n \omega_j^2}\]

Parameters:

• arg (str or barray) – argument to create the subset.
• rescale_weights (bool) – whether to rescale the sample weights.

Returns:

New data set.

Return type:

DataSet

to_backend(backend)

Initialize this class in a different backend.

Parameters:backend (Backend) – new backend. Returns:This class in the new backend.

to_records()

Convert this class into a numpy.ndarray object.

Returns:This object as a a numpy.ndarray object. Return type:numpy.ndarray