mazieres

xmllint --xpath '//*[@PostTypeId="1"]/@Tags' Posts.xml | sed 's/" Tags="/\n/g' | grep 'machine-learning' | sed 's/<\|>/;/g'

mazieres

def ngrams(sequence, depth):
    seq = '^' + sequence + '*'
    res = []
    while depth > 0:
        i, j = 0, depth
        while j <= len(seq):
            res.append(seq[i:j])
            i += 1
            j += 1
        depth -= 1

mazieres

# This script implements the methodology described in Chap. 7 of
# Mateos, Pablo. "Names, Ethnicity and Populations". Springer, 2014.
# For bipartite (fornames, surnames) graph projection for unsupervised
# learning of names ethnicity.
# It uses Louvain instead of FastCommunity however.
# Works fine with pypy.
# 
# by Antoine Mazières (http://mazier.es ; {github|twitter}@mazieres)
# Cortext Lab -- http://www.cortext.net/

mazieres

import unittest
import networkx as nx

class TestExtract(unittest.TestCase):

    def test_graph_from_adj_mat(self):
        X = np.array([
            [(0,), (1,), (2,)],
            [(1,), (0,), (0,)],
            [(2,), (0,), (0,)]], dtype=[('weight', '<i8')])

mazieres

import unittest

class TestExtract(unittest.TestCase): 
    def test_mk_sparse(self):
        # https://en.wikipedia.org/wiki/Sparse_matrix#Dictionary_of_keys_.28DOK.29
        raw = {'A': {'x', 'y', 'z'}, 'B': {'w', 'y'}}
        expected = {('A', 'x'): 1, ('B', 'y'): 1, ('A', 'z'): 1, ('A', 'y'): 1, ('B', 'w'): 1}
        tested = mk_sparse(raw)
        msg = '\nExpected:\n{}\nGot:\n{}'.format(expected, tested)
        self.assertEqual(expected, tested, msg=msg)

mazieres

def symmetrize(a):
    # http://stackoverflow.com/a/2573982/1565438
    return a + a.T - numpy.diag(a.diagonal())

mazieres

def point_inside_polygon(coord, poly):
    '''
    http://www.ariel.com.au/a/python-point-int-poly.html
    '''
    x, y = coord
    n = len(poly)
    inside =False
    p1x,p1y = poly[0]
    for i in range(n+1):
        p2x,p2y = poly[i % n]

mazieres

import unittest

class TestExtract(unittest.TestCase):

    def test_adjacency_matrix(self):
        X = np.array([
            [1, 8, 3],
            [5, 0, 0],
            [0, 4, 2]])
        tested = adjacency_matrix(X)

mazieres

import pandas as pd
import numpy as np


def wannabe_projection(df):
    '''
    https://stats.stackexchange.com/questions/142132/is-this-a-valid-method-for-unipartite-projection-of-a-bipartite-graph
    '''
    n_samples = df.shape[0]
    res = np.zeros((n_samples, n_samples))

mazieres

#!/usr/bin/env python
# by @mazieres for cortext.fr

import sqlite3
import sys
import os

from collections import defaultdict

# PATH to the DB downloaded from cortext