Nature-inspired algorithms for Association Rule Mining
Project description
NiaARM - A minimalistic framework for numerical association rule mining.
General outline of the framework
NiaARM is a framework for Association Rule Mining based on nature-inspired algorithms for optimization. The framework is written fully in Python and runs on all platforms. NiaARM allows users to preprocess the data in a transaction database automatically, to search for association rules and provide a pretty output of the rules found. This framework also supports numerical and real-valued types of attributes besides the categorical ones. Mining the association rules is defined as an optimization problem, and solved using the nature-inspired algorithms that come from the related framework called NiaPy.
Detailed insights
The current version includes (but is not limited to) the following functions:
- loading datasets in CSV format,
- preprocessing of data,
- searching for association rules,
- providing output of mined association rules,
- generating statistics about mined association rules.
Installation
pip3
Install NiaARM with pip3:
pip3 install niaarm
Usage
Basic example
from niaarm import NiaARM, Dataset
from niapy.algorithms.basic import DifferentialEvolution
from niapy.task import Task, OptimizationType
# load and preprocess the dataset from csv
data = Dataset("datasets/Abalone.csv")
# Create a problem:::
# dimension represents the dimension of the problem;
# features represent the list of features, while transactions depicts the list of transactions
# the following 4 elements represent weights (support, confidence, coverage, shrinkage)
# None defines that criteria are omitted and are, therefore, excluded from the fitness function
problem = NiaARM(data.dimension, data.features, data.transactions, alpha=1.0, beta=1.0)
# build niapy task
task = Task(problem=problem, max_iters=30, optimization_type=OptimizationType.MAXIMIZATION)
# use Differential Evolution (DE) algorithm from the NiaPy library
# see full list of available algorithms: https://github.com/NiaOrg/NiaPy/blob/master/Algorithms.md
algo = DifferentialEvolution(population_size=50, differential_weight=0.5, crossover_probability=0.9)
# run algorithm
best = algo.run(task=task)
# sort rules
problem.sort_rules()
# export all rules to csv
problem.export_rules('output.csv')
For a full list of examples see the examples folder.
Command line interface
niaarm -h
usage: niaarm [-h] -i INPUT_FILE [-o OUTPUT_FILE] -a ALGORITHM [-s SEED]
[--max-evals MAX_EVALS] [--max-iters MAX_ITERS] [--alpha ALPHA]
[--beta BETA] [--gamma GAMMA] [--delta DELTA] [--log]
[--show-stats]
Perform ARM, output mined rules as csv, get mined rules' statistics
options:
-h, --help show this help message and exit
-i INPUT_FILE, --input-file INPUT_FILE
Input file containing a csv dataset
-o OUTPUT_FILE, --output-file OUTPUT_FILE
Output file for mined rules
-a ALGORITHM, --algorithm ALGORITHM
Algorithm to use (niapy class name, e. g.
DifferentialEvolution)
-s SEED, --seed SEED Seed for the algorithm's random number generator
--max-evals MAX_EVALS
Maximum number of fitness function evaluations
--max-iters MAX_ITERS
Maximum number of iterations
--alpha ALPHA Alpha parameter. Default 0
--beta BETA Beta parameter. Default 0
--gamma GAMMA Gamma parameter. Default 0
--delta DELTA Delta parameter. Default 0
--log Enable logging of fitness improvements
--show-stats Display stats about mined rules
Note: The CLI script can also run as a python module (python -m niaarm ...
)
Reference Papers:
Ideas are based on the following research papers:
[1] I. Fister Jr., A. Iglesias, A. Gálvez, J. Del Ser, E. Osaba, I Fister. Differential evolution for association rule mining using categorical and numerical attributes In: Intelligent data engineering and automated learning - IDEAL 2018, pp. 79-88, 2018.
[2] I. Fister Jr., V. Podgorelec, I. Fister. Improved Nature-Inspired Algorithms for Numeric Association Rule Mining. In: Vasant P., Zelinka I., Weber GW. (eds) Intelligent Computing and Optimization. ICO 2020. Advances in Intelligent Systems and Computing, vol 1324. Springer, Cham.
[3] I. Fister Jr., I. Fister A brief overview of swarm intelligence-based algorithms for numerical association rule mining. arXiv preprint arXiv:2010.15524 (2020).
License
This package is distributed under the MIT License. This license can be found online at http://www.opensource.org/licenses/MIT.
Disclaimer
This framework is provided as-is, and there are no guarantees that it fits your purposes or that it is bug-free. Use it at your own risk!
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.