-
Notifications
You must be signed in to change notification settings - Fork 0
/
quantile.py
264 lines (205 loc) · 9.4 KB
/
quantile.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Tue Mar 9 15:56:41 2021
@author: alessandro
"""
import numpy as np
from collections.abc import Iterable
class Quantile(object):
"""
A class to create and handle a quantile representation.
A quantile can be created either by data or by simple quantile features
(as number of bins or extension)
...
Attributes
----------
None
Methods
-------
set_by_data(data, nbins, mode='constant', center='half', minvals=1):
Creates a quantile representation by a list or numpy array.
data: list or numpy array
nbins: the number of bins
mode: ['constant', 'frequence'] sets the bins either equally
distributed ('constant') or by keeping the same amount of data
into each bin ('frequence'). Default: 'constant'
center: ['half', 'mean'] sets the bin center either in the half
('half') or on the mean of data containedin that bin ('mean').
Default: 'half'
minvals: applies only in case of center='mean'. If the number of data
points contained in the bin is less than minvals, the relative
center is given by 'half'. Default: 1
set_by_bins(bin_ticks, bin_centers=None):
Creates a quantile representation by an array of bin ticks (the number
of bins is bin_ticks - 1) and, eventually, by an array describing the
bin centers.
bin_ticks: an array containing the bin ticks. Bin ticks are the values
defining the inital point of every bin and the last closing
point. The number of ticks is given by bin_ticks - 1.
bin_centers (optional): an array containing the bin centers. Default:
None
set_by_minmax(minmax, nbins):
Creates a quantile representation by setting the min and max value, and
the number of bins.
minmax: is a tuple containing the min and the max value of the quantile
representation.
nbins: is the total number of bins.
getbin(x): given a value or an array x, getbin returns the related bin or
bins value. The bin value is an integer from 0 to bin max.
getcenter(binvalue): given a bin value or an array binvalue, getcenter
returns the related center(s). It is affected by the center feature
['half', 'mean']
Usage
-----
q = Quantile()
data = np.random.randn(2, 1000)
q.set_by_data(data, 40, mode='constant', center='half', minvals=10)
"""
def __init__(self):
self.nbins = None
self.data = None
self.bin_ticks = None
self.bin_centers = None
self.mode = None
self.center = None
self.source = None
self.minmax = None
self.minvals = None
def set_by_data(self, data, nbins, mode='constant', center='half', minvals=1):
'''
Creates a quantile representation by a list or numpy array.
Parameters:
data: list or numpy array
nbins: the number of bins
mode: ['constant', 'frequence'] sets the bins either equally
distributed ('constant') or by keeping the same amount of data
into each bin ('frequence'). Default: 'constant'
center: ['half', 'mean'] sets the bin center either in the half
('half') or on the mean of data containedin that bin ('mean').
Default: 'half'
minvals: applies only in case of center='mean'. If the number of data
points contained in the bin is less than minvals, the relative
center is given by 'half'. Default: 1
'''
self.source = 'data'
self.nbins = nbins
self.mode = mode
self.minvals = minvals
if self.mode not in ['constant', 'frequence']:
raise ValueError('mode is wrong')
self.center = center
if self.center not in ['half', 'mean']:
raise ValueError('center is wrong')
self.data = np.array(data)
datat = self.data.copy().flatten()
datat = np.sort(datat)
self.bin_ticks = np.zeros(self.nbins + 1)
self.bin_centers = np.zeros(self.nbins)
if mode == 'frequence':
for k in np.arange(self.nbins):
fr = int(k * len(datat) / self.nbins)
to = int((k + 1) * len(datat) / self.nbins)
self.bin_ticks[k] = datat[fr]
if k == self.nbins - 1:
self.bin_ticks[k+1] = datat[-1]
if center == 'half':
self.bin_centers[k] = (self.bin_ticks[k] + self.bin_ticks[k]+1) / 2
elif center == 'mean':
self.bin_centers[k] = datat[fr:to].mean()
elif mode == 'constant':
self.bin_ticks = np.linspace(datat[0], datat[-1], num=nbins+1)
if center == 'half':
self.bin_centers = (self.bin_ticks[1:] + self.bin_ticks[:-1]) / 2
elif center == 'mean':
for k in np.arange(self.nbins):
ind = np.where(np.logical_and(datat>=self.bin_ticks[k], datat<=self.bin_ticks[k+1]))[0]
if len(ind) < minvals:
self.bin_centers[k] = (self.bin_ticks[k] + self.bin_ticks[k+1]) / 2
else:
self.bin_centers[k] = datat[ind].mean()
self.minmax = (self.bin_ticks[0], self.bin_ticks[-1])
def set_by_bins(self, bin_ticks, bin_centers=None):
'''
Creates a quantile representation by an array of bin ticks (the number
of bins is bin_ticks - 1) and, eventually, by an array describing the
bin centers.
Parameters:
bin_ticks: an array containing the bin ticks. Bin ticks are the values
defining the inital point of every bin and the last closing
point. The number of ticks is given by bin_ticks - 1.
bin_centers (optional): an array containing the bin centers. Default:
None
'''
self.source = 'bins'
self.nbins = len(bin_ticks) - 1
self.bin_ticks = np.array(bin_ticks)
if bin_centers is None:
self.bin_centers = (self.bin_ticks[1:] + self.bin_ticks[:-1]) / 2
else:
bin_centers = np.array(bin_centers)
if len(bin_centers) != len(bin_ticks) - 1:
raise ValueError('bin_centers len must be equal to bin_ticks len minus 1')
found = False
for k, bc in enumerate(bin_centers):
if (bc < self.bin_ticks[k]) or (bc > self.bin_ticks[k+1]):
found = True
if found:
raise ValueError('Each element in bin_centers must be inside the relative interval given by bin_ticks')
self.bin_centers = bin_centers
self.minmax = (self.bin_ticks[0], self.bin_ticks[-1])
def set_by_minmax(self, minmax, nbins):
'''
Creates a quantile representation by setting the min and max value, and
the number of bins.
Parameters:
minmax: is a tuple containing the min and the max value of the quantile
representation.
nbins: is the total number of bins.
'''
self.source = 'minmax'
self.nbins = nbins
self.bin_ticks = np.linspace(minmax[0], minmax[1], num=nbins+1)
self.bin_centers = (self.bin_ticks[1:] + self.bin_ticks[:-1]) / 2
self.minmax = (self.bin_ticks[0], self.bin_ticks[-1])
def getbin(self, x):
'''
given a value or an array x, getbin returns the related bin or
bins value. The bin value is an integer from 0 to bin max.
Parameters:
x (float or array): A data point or array
Returns:
binvalue (int or array): The related bin value(s)
'''
if isinstance(x, (int, float)):
binvalue = (x > self.bin_ticks[:-1]).sum()
if binvalue == 0:
binvalue = 1
return binvalue - 1
else:
x = np.array(x)
binvalue = np.zeros_like(x).astype(int)
for k, y in enumerate(x):
binvalue[k] = (y > self.bin_ticks[:-1]).sum()
if binvalue[k] == 0:
binvalue[k] = 1
binvalue[k] -= 1
return binvalue
def getcenter(self, binvalue):
'''
given a bin value or an array binvalue, getcenter
returns the related center(s). It is affected by the center feature
['half', 'mean']
Parameters:
binvalue (int or array): The bin value(s)
Returns:
centers (float or array): The related center(s)
'''
if isinstance(binvalue, Iterable):
binvalue = np.array(binvalue)
centers = np.zeros_like(binvalue).astype(float)
for k, y in enumerate(binvalue):
centers[k] = self.bin_centers[y]
return centers
else:
return self.bin_centers[binvalue]