Coverage for biobb_ml/clustering/k_means_coefficient.py: 85%
93 statements
« prev ^ index » next coverage.py v7.5.1, created at 2024-05-07 09:39 +0000
1#!/usr/bin/env python3
3"""Module containing the KMeansCoefficient class and the command line interface."""
4import argparse
5import pandas as pd
6import numpy as np
7from biobb_common.generic.biobb_object import BiobbObject
8from sklearn.preprocessing import StandardScaler
9from biobb_common.configuration import settings
10from biobb_common.tools import file_utils as fu
11from biobb_common.tools.file_utils import launchlogger
12from biobb_ml.clustering.common import check_input_path, check_output_path, getHeader, getIndependentVars, getIndependentVarsList, hopkins, getWCSS, get_best_K, getGap, getSilhouetthe, plotKmeansTrain
15class KMeansCoefficient(BiobbObject):
16 """
17 | biobb_ml KMeansCoefficient
18 | Wrapper of the scikit-learn KMeans method.
19 | Clusters a given dataset and calculates best K coefficient. Visit the `KMeans documentation page <https://scikit-learn.org/stable/modules/generated/sklearn.cluster.KMeans.html>`_ in the sklearn official website for further information.
21 Args:
22 input_dataset_path (str): Path to the input dataset. File type: input. `Sample file <https://github.com/bioexcel/biobb_ml/raw/master/biobb_ml/test/data/clustering/dataset_k_means_coefficient.csv>`_. Accepted formats: csv (edam:format_3752).
23 output_results_path (str): Table with WCSS (elbow method), Gap and Silhouette coefficients for each cluster. File type: output. `Sample file <https://github.com/bioexcel/biobb_ml/raw/master/biobb_ml/test/reference/clustering/ref_output_results_k_means_coefficient.csv>`_. Accepted formats: csv (edam:format_3752).
24 output_plot_path (str) (Optional): Path to the elbow method and gap statistics plot. File type: output. `Sample file <https://github.com/bioexcel/biobb_ml/raw/master/biobb_ml/test/reference/clustering/ref_output_plot_k_means_coefficient.png>`_. Accepted formats: png (edam:format_3603).
25 properties (dic - Python dictionary object containing the tool parameters, not input/output files):
26 * **predictors** (*dict*) - ({}) Features or columns from your dataset you want to use for fitting. You can specify either a list of columns names from your input dataset, a list of columns indexes or a range of columns indexes. Formats: { "columns": ["column1", "column2"] } or { "indexes": [0, 2, 3, 10, 11, 17] } or { "range": [[0, 20], [50, 102]] }. In case of mulitple formats, the first one will be picked.
27 * **max_clusters** (*int*) - (6) [1~100|1] Maximum number of clusters to use by default for kmeans queries.
28 * **random_state_method** (*int*) - (5) [1~1000|1] Determines random number generation for centroid initialization.
29 * **scale** (*bool*) - (False) Whether or not to scale the input dataset.
30 * **remove_tmp** (*bool*) - (True) [WF property] Remove temporal files.
31 * **restart** (*bool*) - (False) [WF property] Do not execute if output files exist.
33 Examples:
34 This is a use example of how to use the building block from Python::
36 from biobb_ml.clustering.k_means_coefficient import k_means_coefficient
37 prop = {
38 'predictors': {
39 'columns': [ 'column1', 'column2', 'column3' ]
40 },
41 'max_clusters': 3
42 }
43 k_means_coefficient(input_dataset_path='/path/to/myDataset.csv',
44 output_results_path='/path/to/newTable.csv',
45 output_plot_path='/path/to/newPlot.png',
46 properties=prop)
48 Info:
49 * wrapped_software:
50 * name: scikit-learn KMeans
51 * version: >=0.24.2
52 * license: BSD 3-Clause
53 * ontology:
54 * name: EDAM
55 * schema: http://edamontology.org/EDAM.owl
57 """
59 def __init__(self, input_dataset_path, output_results_path,
60 output_plot_path=None, properties=None, **kwargs) -> None:
61 properties = properties or {}
63 # Call parent class constructor
64 super().__init__(properties)
65 self.locals_var_dict = locals().copy()
67 # Input/Output files
68 self.io_dict = {
69 "in": {"input_dataset_path": input_dataset_path},
70 "out": {"output_results_path": output_results_path, "output_plot_path": output_plot_path}
71 }
73 # Properties specific for BB
74 self.predictors = properties.get('predictors', {})
75 self.max_clusters = properties.get('max_clusters', 6)
76 self.random_state_method = properties.get('random_state_method', 5)
77 self.scale = properties.get('scale', False)
78 self.properties = properties
80 # Check the properties
81 self.check_properties(properties)
82 self.check_arguments()
84 def check_data_params(self, out_log, err_log):
85 """ Checks all the input/output paths and parameters """
86 self.io_dict["in"]["input_dataset_path"] = check_input_path(self.io_dict["in"]["input_dataset_path"], "input_dataset_path", out_log, self.__class__.__name__)
87 self.io_dict["out"]["output_results_path"] = check_output_path(self.io_dict["out"]["output_results_path"], "output_results_path", False, out_log, self.__class__.__name__)
88 if self.io_dict["out"]["output_plot_path"]:
89 self.io_dict["out"]["output_plot_path"] = check_output_path(self.io_dict["out"]["output_plot_path"], "output_plot_path", True, out_log, self.__class__.__name__)
91 @launchlogger
92 def launch(self) -> int:
93 """Execute the :class:`KMeansCoefficient <clustering.k_means_coefficient.KMeansCoefficient>` clustering.k_means_coefficient.KMeansCoefficient object."""
95 # check input/output paths and parameters
96 self.check_data_params(self.out_log, self.err_log)
98 # Setup Biobb
99 if self.check_restart():
100 return 0
101 self.stage_files()
103 # load dataset
104 fu.log('Getting dataset from %s' % self.io_dict["in"]["input_dataset_path"], self.out_log, self.global_log)
105 if 'columns' in self.predictors:
106 labels = getHeader(self.io_dict["in"]["input_dataset_path"])
107 skiprows = 1
108 else:
109 labels = None
110 skiprows = None
111 data = pd.read_csv(self.io_dict["in"]["input_dataset_path"], header=None, sep="\\s+|;|:|,|\t", engine="python", skiprows=skiprows, names=labels)
113 # the features are the predictors
114 predictors = getIndependentVars(self.predictors, data, self.out_log, self.__class__.__name__)
115 fu.log('Predictors: [%s]' % (getIndependentVarsList(self.predictors)), self.out_log, self.global_log)
117 # Hopkins test
118 H = hopkins(predictors)
119 fu.log('Performing Hopkins test over dataset. H = %f' % H, self.out_log, self.global_log)
121 # scale dataset
122 if self.scale:
123 fu.log('Scaling dataset', self.out_log, self.global_log)
124 scaler = StandardScaler()
125 predictors = scaler.fit_transform(predictors)
127 # calculate wcss for each cluster
128 fu.log('Calculating Within-Clusters Sum of Squares (WCSS) for each %d clusters' % self.max_clusters, self.out_log, self.global_log)
129 wcss = getWCSS('kmeans', self.max_clusters, predictors)
131 # wcss table
132 wcss_table = pd.DataFrame(data={'cluster': np.arange(1, self.max_clusters + 1), 'WCSS': wcss})
133 fu.log('Calculating WCSS for each cluster\n\nWCSS TABLE\n\n%s\n' % wcss_table.to_string(index=False), self.out_log, self.global_log)
135 # get best cluster elbow method
136 best_k, elbow_index = get_best_K(wcss)
137 fu.log('Optimal number of clusters according to the Elbow Method is %d' % best_k, self.out_log, self.global_log)
139 # calculate gap
140 best_g, gap = getGap('kmeans', predictors, nrefs=5, maxClusters=(self.max_clusters + 1))
142 # gap table
143 gap_table = pd.DataFrame(data={'cluster': np.arange(1, self.max_clusters + 1), 'GAP': gap['gap']})
144 fu.log('Calculating Gap for each cluster\n\nGAP TABLE\n\n%s\n' % gap_table.to_string(index=False), self.out_log, self.global_log)
146 # log best cluster gap method
147 fu.log('Optimal number of clusters according to the Gap Statistics Method is %d' % best_g, self.out_log, self.global_log)
149 # calculate silhouette
150 silhouette_list, s_list = getSilhouetthe(method='kmeans', X=predictors, max_clusters=self.max_clusters, random_state=self.random_state_method)
152 # silhouette table
153 silhouette_table = pd.DataFrame(data={'cluster': np.arange(1, self.max_clusters + 1), 'SILHOUETTE': silhouette_list})
154 fu.log('Calculating Silhouette for each cluster\n\nSILHOUETTE TABLE\n\n%s\n' % silhouette_table.to_string(index=False), self.out_log, self.global_log)
156 # get best cluster silhouette method
157 key = silhouette_list.index(max(silhouette_list))
158 best_s = s_list.__getitem__(key)
159 fu.log('Optimal number of clusters according to the Silhouette Method is %d' % best_s, self.out_log, self.global_log)
161 # save results table
162 results_table = pd.DataFrame(data={'method': ['elbow', 'gap', 'silhouette'], 'coefficient': [wcss[elbow_index], max(gap['gap']), max(silhouette_list)], 'clusters': [best_k, best_g, best_s]})
163 fu.log('Gathering results\n\nRESULTS TABLE\n\n%s\n' % results_table.to_string(index=False), self.out_log, self.global_log)
164 fu.log('Saving results to %s' % self.io_dict["out"]["output_results_path"], self.out_log, self.global_log)
165 results_table.to_csv(self.io_dict["out"]["output_results_path"], index=False, header=True, float_format='%.3f')
167 # wcss plot
168 if self.io_dict["out"]["output_plot_path"]:
169 fu.log('Saving methods plot to %s' % self.io_dict["out"]["output_plot_path"], self.out_log, self.global_log)
170 plot = plotKmeansTrain(self.max_clusters, wcss, gap['gap'], silhouette_list, best_k, best_g, best_s)
171 plot.savefig(self.io_dict["out"]["output_plot_path"], dpi=150)
173 # Copy files to host
174 self.copy_to_host()
176 self.tmp_files.extend([
177 self.stage_io_dict.get("unique_dir")
178 ])
179 self.remove_tmp_files()
181 self.check_arguments(output_files_created=True, raise_exception=False)
183 return 0
186def k_means_coefficient(input_dataset_path: str, output_results_path: str, output_plot_path: str = None, properties: dict = None, **kwargs) -> int:
187 """Execute the :class:`KMeansCoefficient <clustering.k_means_coefficient.KMeansCoefficient>` class and
188 execute the :meth:`launch() <clustering.k_means_coefficient.KMeansCoefficient.launch>` method."""
190 return KMeansCoefficient(input_dataset_path=input_dataset_path,
191 output_results_path=output_results_path,
192 output_plot_path=output_plot_path,
193 properties=properties, **kwargs).launch()
196def main():
197 """Command line execution of this building block. Please check the command line documentation."""
198 parser = argparse.ArgumentParser(description="Wrapper of the scikit-learn KMeans method.", formatter_class=lambda prog: argparse.RawTextHelpFormatter(prog, width=99999))
199 parser.add_argument('--config', required=False, help='Configuration file')
201 # Specific args of each building block
202 required_args = parser.add_argument_group('required arguments')
203 required_args.add_argument('--input_dataset_path', required=True, help='Path to the input dataset. Accepted formats: csv.')
204 required_args.add_argument('--output_results_path', required=True, help='Table with WCSS (elbow method), Gap and Silhouette coefficients for each cluster. Accepted formats: csv.')
205 parser.add_argument('--output_plot_path', required=False, help='Path to the elbow and gap methods plot. Accepted formats: png.')
207 args = parser.parse_args()
208 args.config = args.config or "{}"
209 properties = settings.ConfReader(config=args.config).get_prop_dic()
211 # Specific call of each building block
212 k_means_coefficient(input_dataset_path=args.input_dataset_path,
213 output_results_path=args.output_results_path,
214 output_plot_path=args.output_plot_path,
215 properties=properties)
218if __name__ == '__main__':
219 main()