criteria-performance 1.3.5

A Python package for performance criteria visualization

Performance Criteria

criteria_performance is a Python package that allows you to calculate and visualize performance criteria from CSV data and DataFrames. This module facilitates the evaluation of classification models by calculating essential metrics such as recall, precision, false positive rate (FPR), and false negative rate (FNR), as well as generating ROC, P-R, and DET curves.

Installation

You can install the package via pip:

pip install criteria-performance 

or

pip install criteria_performance 

Usage

Loading Data

To use the module, you need to provide a URL or path to a CSV file containing at least two columns, or a DataFrame:

• The first column must contain the classes (1 for positive, -1 for negative).
• The second column must contain the prediction scores.

Example CSV

Example of a CSV file structure compatible with the module:

Class	Score
1	0.9
1	0.8
-1	0.4
-1	0.1

Example

Here is an example of using the PerformanceCriteria class:

from criteria import PerformanceCriteria

# Replace 'path_to_your_file.csv' with the path to your CSV file
performance = PerformanceCriteria('path_to_your_file.csv')

# Access metrics
print("Recall:", performance.get_recall())
print("Precision:", performance.get_precision())
print("False Positives (FP):", performance.get_fp())
print("False Negatives (FN):", performance.get_fn())

Visualization

The module includes methods to display performance criteria graphs:

• dispROC(): Displays the ROC curve.
• dispPR(): Displays the Precision-Recall curve.
• dispDET(): Displays the DET curve.
• displaygraphe(): Displays all graphs in a single figure.

Example:

performance.displaygraphe()
performance.show()  # Displays the figure

Calculated Metrics

The module calculates and returns the following metrics:

• PPV (Positive Predictive Value): Positive predictive value.
• NPV (Negative Predictive Value): Negative predictive value.
• False Positives (FP): Number of false positives.
• True Positives (TP): Number of true positives.
• False Negatives (FN): Number of false negatives.
• False Negative Rate (FNR): False negative rate.
• False Positive Rate (FPR): False positive rate.
• Recall: Sensitivity or true positive rate.
• Precision: True positives divided by the total predicted positives.

Practical Tools

Here are two useful functions included in the module:

asarray2D(arrayA, arrayB)

Combines two arrays into a 2D array.

Usage:

import numpy as np
from criteria import asarray2D

arrayA = np.array([1, 2, 3])
arrayB = np.array([0.9, 0.8, 0.7])

combined_array = asarray2D(arrayA, arrayB)
print(combined_array)

asDataFrame(array)

Converts a 2D array into a Pandas DataFrame.

Usage:

from criteria import asDataFrame
import numpy as np

array = np.array([[1, 0.9], [1, 0.8], [-1, 0.4]])
df = asDataFrame(array)
print(df)

from criteria import PerformanceCriteria,Opentxt

data = Opentxt("Score_Sys_1.txt")
criter = PerformanceCriteria(data)

Authors

Developed by Olanda-Eyiba Chantry - [e-mail]

License

This project is licensed under the MIT License.