epave

def extract_publish_month(col):
    from sklearn.pipeline import Pipeline
    from sklearn.preprocessing import OneHotEncoder
    import numpy as np
    
    class MonthExtractor():
        
        def __init__(self):
            pass
        

epave

def tfidf_profarea(col, min_df=100, profarea_only=True):
    from sklearn.feature_extraction.text import TfidfTransformer
    from sklearn.feature_extraction.text import CountVectorizer
    from sklearn.pipeline import Pipeline
    
    class IdExtractor():
        
        def __init__(self, field='profarea_id'):
            self._field=field
        

epave

import os, subprocess,sys
def lifx_run(token, command, params):
    url = '"https://api.lifx.com/v1beta1/lights/all/' + command + '"'
    if command.startswith('effects'):
        data = "color="+params
        method = 'POST'
    else:
        if command == 'power':
            data = "state="+params
        elif command == 'color':

epave

def read_adj_rand_score(fname):
    import scipy.sparse as sp
    from scipy.misc import comb
    def comb2(n):
        # the exact version is faster for k == 2: use it by default globally in
        # this module instead of the float approximate variant
        return comb(n, 2, exact=1)
    mat = sp.lil_matrix((5000,5000), dtype=float)
    with open(fname) as f:
        lines = f.readlines()

epave

def my_scores(labels_one, labels_two):
    vprod = len(intersect1d(labels_one.nonzero()[0], labels_two.nonzero()[0]))
    if vprod < 100:
        return 0.0, 0.0
    n_samples = labels_one.shape[0]
    nnz_one = labels_one.nnz
    nnz_two = labels_two.nnz
    contingency = array([[labels_one.shape[0] - nnz_one - nnz_two + vprod, nnz_one - vprod],[nnz_two - vprod, vprod]], dtype='float')
    contingency_sum = labels_one.shape[0]#np.sum(contingency)
    pi = array([labels_one.shape[0] - nnz_two, nnz_two])#np.sum(contingency, axis=1)