Third course reasearch project on developing the way to cluster bank clients from date, time and coordinates in their transaction history
Project description
# Geographical Transctions clustering algorithm
Name of the module stands for geographical transactions clustering. This module is an implementation of the method, developed for the third course project in HSE University. It takes dataframe with clients transactions history of the specified format and returns list of clusters.
For the record, it was intended to be for public usage in this form, as it is a research project seeking to find a way to deal with the described problem
## Installation
Run the following to install:
‘’’python
pip install geot_cluster’’’
## Usage
Before using make sure, that your dataset corresponds with requirements. Csv file must contain the following columns in order to work correctly
user_id : string type, example: “423156821”
event_dt : string type, example: “20190312”
event_time: string type, example: “2019-03-12 06:24:00.279”
lattitude : float type, example: 49.862621
longtitude: see lattitude
‘’’python
import pandas as pd import numpy as np import markov_clustering as mc import matplotlib.pyplot as plt import math import pytz import folium import os.path import networkx as nx
from haversine import haversine, Unit from collections import Counter from datetime import datetime from timezonefinder import TimezoneFinder from IPython.display import clear_output
import geotrans_cluster
path = [path to file with data] data, names = data_load(path)
%matplotlib notebook base = [path to the folder, where to store libs with information about clients]
archivate = True libs= True graph_f = True cluster_f = True
- if(archivate):
archivate_maps(data, names, levels=4)
- if(libs):
lib = graph_preparation(data, names, base) prob_lib = znakomstvo_by_lib(lib,data)
lib, prob_lib = load_libs(base = base)
- if(graph_f):
graph = graph_forming(lib, prob_lib, treshold=0.9)
- if(cluster_f):
result = mc.run_mcl(graph,pruning_threshold=0.7, inflation=2,expansion=2) clusters = mc.get_clusters(result)
clust_0 = clusters_to_ids(lib=lib, prob_lib=prob_lib, clusters = clusters, number = 0) maps = get_cluster_maps(data = data, clust = clust_0) print(“Number of clusters”, len(clusters))
plt.figure(figsize=(10,10)) mc.drawing.draw_graph(result, clusters, edge_color=”red”,node_size=15,width = 1, with_labels=True, font_size = 8)
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.
Source Distribution
Built Distribution
Hashes for geot_cluster-0.9.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | c9269035202a6fee1e0b5ee9df8ac6ca98364981823d5e07f3039f296e813e22 |
|
MD5 | 5b8ea7c217f495fe88577b9991bc1b69 |
|
BLAKE2b-256 | ee6956900d258cd1b2cdb5a972e44d9c78fe1d61a9b9ec5f96ad0ec5187ef7e6 |