fg_cluster_executer.py

import pandas as pd
import umap
from pyclustering.cluster import xmeans

class FGClusterExecuter:
    """
    FanGraphs stats clustering class
    """
    
    @classmethod
    def cluster(cls, csv_file, labels, sep=','):
        labels, data = cls.__load_data(csv_file, labels, sep)
        ndata = cls.__normalize(data)
        arrdata = ndata.as_matrix()
        umap_emb = cls.__exec_umap(arrdata)
        clusters = cls.__exec_xmeans(arrdata)
        return labels, data, umap_emb, clusters
    
    @staticmethod
    def __load_data(csv_file, labels, sep):
        df = pd.read_csv(csv_file, sep=sep)
        labels =df[labels]
        data = df.drop(labels, axis=1)
        return labels, data
    
    @staticmethod
    def __normalize(df):
        df = df.fillna(0)
        ndata = (df - df.mean()) / (df.max() - df.min())
        return ndata.fillna(0)
    
    @staticmethod
    def __exec_umap(array):
        return umap.UMAP().fit_transform(array)
    
    @staticmethod
    def __exec_xmeans(array):
        init_center = xmeans.kmeans_plusplus_initializer(array, 2).initialize()
        xm = xmeans.xmeans(array, init_center, ccore=False)
        xm.process()
        return xm.get_clusters()