aneic-scripts/write_html.py at master · aneic/aneic-scripts · GitHub

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
'''
Generate html file of results
'''
import os
import sys
import pickle
import gzip
from string import Template
import numpy as np
import pandas as pd

sys.path.append('./aneic-core/src')
from aneic import mfm
from aneic import mutual

# Template string for HTML for page
doc_tpl = Template('''\
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
    <link rel="stylesheet" type="text/css" href="layout.css"/>
</head>
<body>
    <h2>$K Clusters</h2>
    <br>
    <table>
$headers
$rows
    </table>
</body>
</html>
''')

# Template for table rows
row_tpl = Template('''\
        <tr>
$cells
        </tr>\
''')

# Template for table cells (cluster descriptions)
desc_tpl = Template('''\
            <td>
$desc
            </td>\
''')

# Template for table cells (individual objects)
cell_tpl = Template('''\
            <td>
                <h3>$name</h3>
                <br>
                <a href="$url">
                    <img src="$url">
                </a>
$desc
            </td>\
''')

# Template for empty cells
cell_empty = '''\
            <td>
            </td>\
'''

def mutual_inf_dz(real, cat, theta, gamma, bins=10):
    mixz, miyz = mutual.MI0dz(real, cat,
                    theta['mu'], theta['l'], theta['rho'], theta['pi'],
                    gamma, bins=bins)
    return pd.concat([mixz, miyz])

def write_html(filename, real, cat, theta, gamma, mi=None, maxft=5,
    base_url='%s.jpg'):
    if not mi is None:
        # get sorted list of top features
        top = mi.order(ascending=False).index
        # get feature counts per cluster
        real_count = pd.DataFrame((np.asarray(~pd.isnull(real))[:,None,:]
                                   * np.asarray(gamma)[:,:,None]).sum(0),
                        index=gamma.columns, columns=real.columns)
        cat_count = pd.DataFrame((np.asarray(cat.fillna(0))[:,None,:]
                                   * np.asarray(gamma)[:,:,None]).sum(0),
                        index=gamma.columns, columns=cat.columns)
    else:
        top = []

    # number of objects and clusters
    N = len(real)
    K = theta['mu'].shape[1]

    # build header strings
    headers = ['        <th>Cluster %d</th>' % (k+1) for k in range(K)]

    # build strings for description cells of each cluster
    if not mi is None:
        descs = []
        for k in range(K):
            indent16 = ' '.join(['']*16)
            desc = '%s<p>Objects: %.1f</p>' \
                    % (indent16, gamma.ix[:,k].sum())
            f = 0
            for ft in top:
                info = ''
                if ft in real.columns:
                    info = '%.2f +/- %.2f (%d)' \
                           % (theta['mu'].ix[ft,k],
                              theta['l'].ix[ft,k]**-0.5,
                              real_count.ix[k,ft])
                if ft in cat.columns.levels[0]:
                    infos = ['%s: %.1f' % p for p in zip(cat[ft].columns,
                                                         cat_count.ix[k,ft])]
                    info = ', '.join(infos)
                if info:
                    desc += '\n%s<p>%s: %s</p>' \
                            % (indent16, ft, info)
                    f +=1
                if f == maxft:
                    break
            descs.append(desc_tpl.substitute(desc=desc))

    # build strings for objects cells of each cluster
    cells = [[] for k in range(K)]
    for k in range(K):
        # find objects in cluster
        msk = np.array(gamma.idxmax(axis=1) == k)
        g = gamma[msk]
        h = (- (g * np.log(g)).sum(axis=1))
        h.sort()
        for nm in h.index:
            url = base_url % nm
            # add gamma to info block
            indent16 = ' '.join(['']*16)
            gstrings = ['%.2f' % g for g in gamma.ix[nm, :]]
            desc = '%s<p>weights: %s</p>' \
                    % (indent16, ', '.join(gstrings))
            # add info on top features if available
            f = 0
            for ft in top:
                val = ''
                if ft in real.columns and pd.notnull(real.ix[nm,ft]):
                    val = '%.2f' % real.ix[nm,ft]
                if ft in cat.columns.levels[0] and pd.notnull(cat.ix[nm,ft]).any():
                    val = '%s' % cat.ix[nm,ft][cat.ix[nm,ft].nonzero()[0]].index[0]
                if val:
                    desc += '\n%s<p>%s: %s (%.2f)</p>' \
                            % (indent16, ft, val, mi[ft])
                    f +=1
                if f == maxft:
                    break
            # add to cells
            cell = cell_tpl.substitute(name=nm,
                                       url=url,
                                       desc=desc)
            cells[k].append(cell)

    # build row strings
    R = max([len(c) for c in cells])
    rows = []
    for r in range(R):
        row = []
        for k in range(K):
            if r < len(cells[k]):
                row.append(cells[k][r])
            else:
                row.append(cell_empty)
        rows.append(row_tpl.substitute(cells='\n'.join(row)))
    if not mi is None:
        rows.insert(0, row_tpl.substitute(cells='\n'.join(descs)))

    # build page string
    doc = doc_tpl.substitute(K = str(K),
                             headers = '\n'.join(headers),
                             rows = '\n'.join(rows))

    # write output
    f = open(filename, 'w')
    f.write(doc)
    f.close()

if __name__ == '__main__':
    # parse command-line args
    from optparse import OptionParser
    usage = "usage: %prog [options] files"
    parser = OptionParser(usage=usage)
    parser.add_option('-o', '--output-dir',
                      metavar="DIR", help="output directory", dest='outdir')
    parser.add_option('-i', '--image-dir',
                      metavar="DIR", help="image directory", dest='imdir')
    (opt, args) = parser.parse_args()

    # find files matching pattern for each supplied argument
    import glob
    files = []
    for pattern in args:
        files += glob.glob(pattern)

    # create output dir if necessary
    import os
    if opt.outdir and files:
        html_dir = opt.outdir
        if not os.path.isdir(html_dir):
            os.makedirs(html_dir)
    else:
        html_dir = os.getcwd()

    im_dir = opt.imdir if opt.imdir else './aneic-core/data/images/'
    base_url = os.path.relpath(im_dir, html_dir) + '/%s.jpg'

    # write out html
    data = None
    for fname in files:
        # load output
        mfm_output = pickle.load(gzip.open(fname, 'r'))
        # get dataset
        real = mfm_output['real']
        cat = mfm_output['cat']
        # get parameters
        theta = mfm_output['theta']
        # get state assignments
        gamma = mfm_output['gamma']
        # get mutual information
        midz = mutual_inf_dz(real, cat, theta, gamma, bins=5)
        # if data is None:
        #     # get path
        #     path = os.getcwd()
        #     # load dataset
        #     opts = mfm_output['opts']
        #     # load data
        #     data = pd.read_csv('%s/%s' %(path, opts['filename']), index_col=1,
        #                        na_values=opts['na_values'])\
        #                .drop(opts['drop_feats'], axis=1)\
        #                .drop(opts['drop_objs'], axis=0)
        #     # split real and cat feats
        #     cat = mfm.convert_indicator(data[data.columns[data.dtypes == np.object]])
        #     real = data[data.columns[data.dtypes == np.float64]]
        # write html
        hname = fname.replace('.pk.gz', '.html').replace('results','html')
        base_path = os.path.relpath('./aneic-core/data/images/', html_dir)
        write_html(hname, real, cat, theta, gamma,
            mi=midz, base_url=base_url)