LPatternIdentification

https://pypi.org/project/LPatternIdentification/

The formal mathematical definition of the l-Pattern Identification Problem is as follows:

Input:

A finite alphabet Σ, two disjoint sets Good, Bad ⊆ Σn of strings and an integer l > 0

Problem question:

Is there a set of P patterns such that: |P| ≤ l and P → (Good, Bad)?

Demonstration:

See demo.ipynb

1. Install:

pip install LPatternIdentification

2. Load:

from LPatternIdentification import feature_set, split_data, get_patterns_from_feature_set, reduce_pattern_set

3. Prepare data:

Sort dataset by class labels

Separate observations into numpy ndarray

Separate labels into list

4. Find set of features

features = feature_set(observations, labels)

5. Split observations by classes

Here, classes are named 'Good' and 'Bad', the 'Good' class being the class of our interest.

split_point, Good, Bad  = split_data(elections_X, elections_y)

6. Return a set of patterns that contain the features and are Good

Patterns = get_patterns_from_feature_set(Good, elections_feature_set)

7. Identify 'L' number of patterns such that all patterns are uniquely Good and not similar to Bad patterns

L = 7
Patterns_identified = reduce_pattern_set(Patterns, Bad, L)