Code Structure

How it works

qp is built upon the functionality of scipy.stats distributions. The base parameterization classes inherit from the SciPy rv_continuous and rv_frozen classes. Each parameterization class must inherit from these classes, so it will have all the functionality of the SciPy classes, plus the additional qp functionality (metadata, normalization functions, etc). In each parameterization class, certain rv_continuous methods such as _pdf() and _cdf() are overridden by new methods specific to that parameterization (see Creating new parameterizations for more details).

Ensembles are then created using one of these parameterization classes. The Ensemble class creates the metadata and objdata dictionaries in addition to the ancil dictionary. It also handles qp specific functionality such as the ability to convert between parameterizations, or write your Ensemble to file.

Organization

qp is split up into two main sections: core for the main functionality and parameterizations for the classes and supporting functions for each parameterization type.

• core

  • Ensemble: the primary code object

  • factory.py: supporting functionality, wraps scipy.stats distributions as qp parameterizations, stores dictionary of all available parameterizations

• parameterizations

  • base.py: contains the base classes that all parameterization classes must inherit from

  • parameterization classes: contain the attributes and functions specific to that parameterization

    • main file: contains the parameterization class and its methods

    • _utils.py file: contains any supporting methods required for the parameterization

Below is a diagram of the current code structure, and some descriptions of the purpose of files and folders.

qp
├── __init__.py
├── _version.py
├── core
│   ├── __init__.py
│   ├── ensemble.py
│   ├── factory.py
│   └── lazy_modules.py
├── metrics # this module contains functions to calculate metrics on qp objects
│   ├── __init__.py
│   ├── array_metrics.py
│   ├── base_metric_classes.py
│   ├── brier.py
│   ├── concrete_metric_classes.py
│   ├── factory.py
│   ├── goodness_of_fit.py
│   ├── metrics.py
│   ├── parallel_metrics.ipynb
│   ├── pit.py
│   ├── point_estimate_metric_classes.py
│   └── util_funcs.py
├── parameterizations # contains code for all parameterizations
│   ├── __init__.py
│   ├── analytic_parameterizations # code for analytic parameterizations
│   │   ├── __init__.py
│   │   ├── mixmod # mixed model Gaussian parameterization
│   │   │   ├── __init__.py
│   │   │   ├── mixmod.py
│   │   │   └── mixmod_utils.py
│   ├── base.py # base classes for all parameterizations
│   ├── hist # histogram parameterization
│   │   ├── __init__.py
│   │   ├── hist.py
│   │   └── hist_utils.py
│   ├── interp # interpolated parameterization
│   │   ├── __init__.py
│   │   ├── interp.py
│   │   └── interp_utils.py
│   ├── packed_interp # interpolated parameterization stored as packed integers
│   │   ├── __init__.py
│   │   ├── packed_interp.py
│   │   └── packing_utils.py
│   ├── quant # quantile parameterization
│   │   ├── __init__.py
│   │   ├── abstract_pdf_constructor.py # base constructor class
│   │   ├── cdf_spline_derivative.py # constructor class
│   │   ├── dual_spline_average.py # constructor class
│   │   ├── piecewise_constant.py # constructor class
│   │   ├── piecewise_linear.py # default constructor class
│   │   ├── quant.py
│   │   └── quant_utils.py
│   ├── sparse_interp # sparse parameterization
│   │   ├── __init__.py
│   │   ├── sparse.py
│   │   ├── sparse_rep.py
│   │   └── sparse_utils.py
│   └── spline # spline parameterization
│       ├── __init__.py
│       ├── spline.py
│       └── spline_utils.py
├── plotting.py # creates plots
├── utils # utility functions used throughout qp
│   ├── __init__.py
│   ├── array.py # performing array operations
│   ├── conversion.py # utilities for converting between parameterizations and unused functions
│   ├── dictionary.py # performing dictionary operations
│   ├── interpolation.py # interpolation functions
└── version.py