laing20333

# coding=utf-8
# Goal: parse house information for each district from websites
# for each district, get 「土地區段位置或建物區門牌」,「建物型態」,「建物現況格局」,「坪數」,「屋齡」,「總價元」,「資料來源」into csv file

# Procedure:
# 1. get the number of page for each district by parsing first html content
# 2. for each district put all html page together, use htmlparser to parse content and save data into file

import sys
import math

laing20333

#coding=utf-8
import distance_fun
import timeit
import numpy as np
import matplotlib.pyplot as plt

t_python = timeit.Timer("clustering.kp_distance(np.array([1, 2, 4, '住', '公寓']), np.array([2, 2, 2, '商', '公寓']), 0.5)" , "import clustering\nimport numpy as np")
time_python = t_python.timeit(1000000)
print 'python code execution time: ' + str(time_python) + 's'

laing20333

def kp_distance(object_x, object_y, Wc):
    ''' Distance function for two objects
        '''
    res = 0.0
    for attr_idx in xrange(0, len(object_x)):
        cur_type = type(object_x[attr_idx])
        if( (cur_type == str) or (cur_type == np.string_) ):
            # categorical attribute
            if (object_x[attr_idx] == object_y[attr_idx]):
                res = res + Wc

laing20333

cimport numpy as np
cimport cython
import numpy as np
def kp_distance(np.ndarray object_x, np.ndarray object_y, float Wc):
    ''' Distance function for two objects
        '''
    cdef float res = 0.0
    cdef unsigned attr_idx
    for attr_idx in xrange(0, len(object_x)):
        cur_type = type(object_x[attr_idx])

laing20333

# coding=utf-8
# Goal:
# parse different city's website
# for each city, get  「土地區段位置或建物區門牌」,「建物型態」,「建物現況格局」,「坪數」,「屋齡」,「總價元」,「資料來源」into csv file

# step:
# 1. get number of page by parsing string
# 2. put all html page together, use htmlparser to get content
# 3. parsing content and save it

laing20333

#include "Cure.h"
#include<iostream>
#include<cmath>
#include<algorithm>
#include<memory.h>

using namespace std;

bool operator< (const struct CLUSTER_PAIR &a, const struct CLUSTER_PAIR &b){
	return (a.distance < b.distance);