poetic.results module¶

Module for processing prediction results.

The results module processes the outputs of prediction results from the Predictor of the predictor module. The functionalities provided include statistical summaries, io, and diagnostic reports.

Examples

The results module’s Diagnostics module can be used as part of the prediction workflow. To use its methods with the Predictor class:

import poetic

pred = poetic.Predictor()
result = pred.predict("This is an example.")
result.run_diagnostics() # One-stop function
result.to_file("<PATH>")

To use without the Predictor class:

import poetic

pred = [0.1, 0.2, 0.3]
result = Diagnostics(predictions=pred)
five_number_summary = result.five_number()

All public methods can be used without the run_diagnostics() method, but they depend on the diagnostic attribute, which the run_diagnostic() method generates.

class poetic.results.Diagnostics(predictions: List[float], sentences: Optional[List[str]] = None)¶

Bases: object

Class for storing and processing prediction results.

Diagnostics is the default base class of Predictions, which is generated by the Predictor class. It can also be used as a standalone class for processing any numeric results and strings stored in lists.

Parameters

predictions (list) – Predictions of poetic scores.
sentences (list, optional) – Sentences associated with the predictions.

predictions¶

Predictions of poetic scores.

Type: list

sentences¶

Sentences associated with the predictions.

Type: list

diagnostics¶

A dictionary of diagnostics statistics, including sentence count, five number summary, and the predictions themselves.

Type: dict

__add__(rhs: poetic.results.Diagnostics) → poetic.results.Diagnostics ¶

Method for + operator.

The + operator concatenates two Diagnostics objects by concatenating all its predictions and diagnostics attributes. If one object’s diagnostics attribute is not None, the run_diagnostics() method will be called on the returned object as well.

Returns: A new Diagnostics as a concatenated result.
Return type: Diagnostics
Raises: TypeError – Error when concatenating non-Diagnostics objects,

__ge__(rhs: poetic.results.Diagnostics) → bool¶

Method for > operator.

The >= operator relies on the mean predictions of each object for comparison, and it is suited for normal or normal-like data.

Returns

Whether the mean predictions of the left-hand-side is: greator or equal.

Return type

bool

__gt__(rhs: poetic.results.Diagnostics) → bool¶

Method for > operator.

The > operator relies on the mean predictions of each object for comparison, and it is suited for normal or normal-like data.

Returns: Whether the mean predictions the left-hand-side is greator.
Return type: bool

__iadd__(rhs: poetic.results.Diagnostics) → poetic.results.Diagnostics ¶

Method for += operator.

The += operator concatenates a new Diagnostics objects to the left-hand-side by concatenating predictions and sentences attributes. If either object’s diagnostics attribute is not None, the run_diagnostics() method will be called on the original object.

Returns: The concatenated Diagnostics object.
Return type: Diagnostics
Raises: TypeError – Error when concatenating non-Diagnostics objects,

__le__(rhs: poetic.results.Diagnostics) → bool¶

Method for > operator.

The <= operator relies on the mean predictions of each object for comparison, and it is suited for normal or normal-like data.

Returns

Whether the mean predictions the left-hand-side is smaller: or equal.

Return type

bool

__lt__(rhs: poetic.results.Diagnostics) → bool¶

Method for < operator.

The < operator relies on the mean predictions of each object for comparison, and it is suited for normal or normal-like data.

Returns: Whether the mean predictions the left-hand-side is smaller.
Return type: bool

__repr__() → str¶

String representation for repr().

This string representation returns a dictionary cast into a string with three attributes of this class: predictions, sentences, and diagnostics.

Returns: String representation of the object.
Return type: str

__str__() → str¶

String representation for str().

This string representation returns a summary of of the object. It will truncate results if there are more than 15 prediction entries. To get a string representation of all the results in a dictionary cast into string, use repr() method.

Returns: String representation of the object.
Return type: str

classmethod five_number(numeric_input: Union[numpy.ndarray, List[float]] = None, **kwargs) → Dict[str, float]¶

Five number summary.

This methods generates five number summary of a given input. The five number summary includes minimum, mean, median, standard deviation, and maximum. This is a class method.

Parameters: numeric_input (numpy.ndarrau, list) – An array like object.
Returns: A dictionary of five number results.
Return type: dict(str, float)

generate_report() → str¶

Generates the diagnostics report in string.

This methods generates a diagnostics report as a string, with Poetic package information, five number summary, and all sentences and their poetic sores.

Returns: A string with diagnostic report.
Return type: str

run_diagnostics() → None¶

Run the diagnostics of the predictions.

This methods generate diagnostics of the predictions, which include sentence count, five number summary, and the sentences themselves.

to_csv(path: str) → None¶

Saves predictions and sentences to a csv file.

This methods saves the results to a csv file. For a plain text diagnostics, please use the to_file() method.

Parameters: path (str) – An string representing the file path.

to_file(path: str) → None¶

Saves diagnostics and predictions to a file.

This methods saves the results to a csv or generates a diagnostics report along with the predictions. The supplied file path’s file extension is used to determine which file to save. If a csv is explicitly desired, to_csv() method can be used. For all file extensions other than csv, a plain text report will be generated.

Parameters: path (str) – An string representing the file path.