uravu.distribution

The storage and manipulation of probability distributions is fundamental to the operation of uravu and Bayesian inference. The Distribution class oversees these operations.

class uravu.distribution.Distribution(samples: Union[List[Union[float, int]], numpy.ndarray], name: str = 'Distribution', random_state: numpy.random._generator.Generator = None)

Bases: object

In addition to storage of the probability distribution, this class allows for some basic analysis, such as determination of normality.

samples

Samples in the distribution.

Type:array_like

name

Distribution name.

Type:str

normal

Are the samples normally distributed?

Type:bool

kde

Kernel density approximation for the distribution.

Type:scipy.stats.kde.gaussian_kde

Parameters:

• samples (array_like) – Sample for the distribution.
• name (str, optional) – A name to identify the distribution. Default is 'Distribution'.
• ci_points (array_like, optional) – The two percentiles at which confidence intervals should be found. Default is [2.5, 97.5] (a 95 % confidence interval).

add_samples(samples: Union[List[Union[float, int]], numpy.ndarray])

Add samples to the distribution.

Parameters:samples (array_like) – Samples to be added to the distribution.

check_normality() → bool

Uses a scipy.stats.normaltest() to evaluate if samples are normally distributed and updates the normal attribute.

ci(ci_points: List[float] = [2.5, 97.5]) → numpy.ndarray

Get the extrema of the confidence intervals of the distribution.

Parameters:ci_points – The confidence interval points to return. Returns:Distribution values at the confidence interval.

con_int(ci_points: List[float] = [2.5, 97.5]) → numpy.ndarray

Get the extrema of the confidence intervals of the distribution.

Parameters:ci_points – The confidence interval points to return. Returns:Distribution values at the confidence interval.

dist_max

Get the value that maximises the distribution. If no kde has been created (for example if the distribution has fewer than 8 values) the median is returned.

Returns:Most likely value.

classmethod from_dict(my_dict: dict) → uravu.distribution.Distribution

Parameters:my_dict – Dictionary description of the distribution. Returns:Distribution object form the dictionary.

logpdf(x: Union[float, List[Union[float, int]], numpy.ndarray]) → Union[float, numpy.ndarray]

Get the natural log probability density function for the distribution.

Parameters:x – Value to return natural log probability of. Returns:Natural log probability.

max

Sample maximum.

Type:return

min

Sample minimum.

Type:return

n

Median value.

Type:return

negative_pdf(x: Union[float, List[Union[float, int]], numpy.ndarray]) → Union[float, numpy.ndarray]

Get the negative of the probability density function for the distribution.

Parameters:x – Value to return negative probability of. Returns:Negative probability.

pdf(x: Union[float, List[Union[float, int]], numpy.ndarray]) → Union[float, numpy.ndarray]

Get the probability density function for the distribution.

Parameters:x – Value to return probability of. Returns:Probability.

s

Standard deviation of the distribution. For a non-normal distribution, this will return None.

Type:return

size

Number of samples in the distribution.

Type:return

to_dict() → dict

Returns:Dictionary description of the object.

v

Standard deviation of the distribution. For a non-normal distribution, this will return None.

Type:return