Bacteria-Farm

Architecture

Description

An intuitive density-based clustering algorithm, which is split into two phases:

Measure of Center Extraction

The model uses a standard clustering algorithm to extract both the measure of center ( typically centroids ) and the proportion of test data in each of the resultant clusters. Moreover, the clustering at this phase can be halted prematurely once these parameters start to converge towards their final values. We have this flexibility as the standard algorithm is essential just for the presentation of those prerequisited parameters to the core model.

Core Model

The core model identifies a set of front runners - data points which sufficiently represent each cluster - and uses them to expand their clusters until the terminating conditions are met. This method is detailed further under the Documents directory.

Target Applications

This model is particularly suited for applications with intent to improve the accuracy ( of clustering ) while accepting a trade-off on the runtime of the model ( as it inherently involves prior clustering ). Additionally, the model provides flexibility with the initial algorithm used ( for dataset-level customizations ) and also features noise specification, which are valuable add-ons for such applications.

Reads

Please scan through the Documents directory for an academic reading.

Model Changes

1. The final version of the model is 'BFFR Sequential and Optimized.py' under the Code directory.
2. It evolved as follows ( all scripts are under the same Code directory )

• BF.py
• BF Improved.py
• BFFR.py
• BFFR Sequential.py
• BFFR Sequential and Optimized.py

Supported Data Formats

1. CSV

Usage

Step 1.Install scikit-learn and Numpy

pip3 install -U sklearn
pip3 install -U numpy

Step 2. Clone this repository to local.

git clone https://github.com/akhil-reddy/Bacteria-Farm
cd Bacteria-Farm
cd Code

Step 3. Run the clustering algorithm

python3 "BFFR Sequential and Optimized.py" <path to the csv file>