HDBScan

class geoanalytics.clustering.HDBScan.HDBScan(dataframe)

Bases: object

About this algorithm

Description:

HDBSCAN (Hierarchical Density-Based Spatial Clustering of Applications with Noise) is an advanced density-based clustering algorithm. It extends DBSCAN by converting it into a hierarchical clustering algorithm and then using a technique to extract a flat clustering based on the stability of clusters. This wrapper enables runtime and memory usage tracking, and allows exporting results to CSV.

Parameters:

• Dataset (pandas DataFrame) must be provided during object initialization.

• Clustering hyperparameters are passed to the run() method.

Attributes:

• df (pd.DataFrame) – The input data with ‘x’, ‘y’ coordinates and features.

• labelsDF (pd.DataFrame) – DataFrame containing ‘x’, ‘y’, and assigned cluster labels.

• startTime, endTime (float) – Runtime tracking.

• memoryUSS, memoryRSS (float) – Memory usage of the clustering process in kilobytes.

Execution methods

Calling from a Python program

import pandas as pd

from geoanalytics.clustering import HDBScan

df = pd.read_csv("input.csv")

hdb = HDBScan(df)

labels_df = hdb.run(min_sample=5, min_cluster_size=10)

hdb.getRuntime()
hdb.getMemoryUSS()
hdb.getMemoryRSS()

hdb.save('HDBScanLabels.csv')

Credits

This implementation was created by Raashika and revised by M.Charan Teja under the guidance of Professor Rage Uday Kiran.

getMemoryRSS()

Prints the memory usage (RSS) of the process in kilobytes.

getMemoryUSS()

Prints the memory usage (USS) of the process in kilobytes.

getRuntime()

Prints the total runtime of the clustering algorithm.

run(min_sample=5, min_cluster_size=10)

Executes HDBScan clustering algorithm.

Parameters:

• min_sample – int, minimum samples required for a dense region (default: 5)

• min_cluster_size – int, minimum size of clusters to extract (default: 10)

Returns:

labelsDF (pd.DataFrame) with columns [‘x’, ‘y’, ‘labels’]

save(outputFileLabels='HDBScanLabels.csv')

Saves the clustering result with labels to a CSV file.

Parameters:

outputFileLabels – str, filename for saving labels (default: ‘HDBScanLabels.csv’)