statistics.py 10.1 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
# Copyright 2018 Markus Scheidgen
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an"AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

"""
A command that generates various statistics.
"""

19
20
21
22
23
from matplotlib import scale as mscale
from matplotlib import transforms as mtransforms
import matplotlib.pyplot as plt
import matplotlib.ticker as ticker
import numpy as np
24
import click
25

26
from .client import client
27
28


29
30
def codes(client, metrics=[]):
    data = client.repo.search(per_page=1, owner='admin', metrics=metrics).response().result
31

32
33
    x_values = sorted([
        code for code, values in data.quantities['code_name'].items()
34
        if code != 'not processed' and values['code_runs'] > 0], key=lambda x: x.lower())
35

36
37
38
39
40
41
42
43
44
45
46
    return data.quantities, x_values, 'code_name', 'code'


def dates(client, metrics=[]):
    data = client.repo.search(per_page=1, owner='admin', metrics=metrics, date_histogram=True).response().result

    x_values = list([
        x for x in data.quantities['date_histogram'].keys()])

    return data.quantities, x_values, 'date_histogram', 'month'

47

48
def error_fig(client):
49
    _, labels, _, _ = codes(client)
50
51
52
53
54
55
56
57
58
59
60

    def code_values(metric='code_runs', **kwargs):
        result = client.repo.search(
            per_page=1,
            owner='admin',
            metrics=[] if metric == 'code_runs' else metric,
            **kwargs).response().result

        return {
            code: values[metric]
            for code, values in result.quantities['code_name'].items()
61
            if code != 'not processed' and (not labels or code in labels) > 0}
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

    # get the data
    all_entries = code_values()
    parser_failure_label = 'parser failure'
    error_types = [
        {'name': parser_failure_label, 'search': dict(system='not processed')},
        {'name': 'failed system classification', 'search': dict(system='unavailable')},
        {'name': 'no basis set available', 'search': dict(basis_set='unavailable')},
        {'name': 'no XC functional available', 'search': dict(xc_functional='unavailable')}
    ]
    errors = {
        error_type['name']: {
            code: failures
            for code, failures in code_values(**error_type['search']).items()}
        for error_type in error_types}
    errors_rates = {
        error_type['name']: {
            code: 0 if all_entries[code] == 0 else failures / all_entries[code]
            for code, failures in code_values(**error_type['search']).items()}
        for error_type in error_types}

    fig, axs = plt.subplots(figsize=(15, 12), dpi=72, nrows=2)

    def draw_error_chart(errors, ax, colors, entries=None, mul=1, scale=0.5):
        n_bars = len(errors) - 1
        leg_colors = list(colors)

        x = np.arange(len(labels))  # the label locations
        width = 0.7 / n_bars  # the width of the bars
        plt.sca(ax)
        plt.xticks(rotation=90)

        if entries is not None:
            ax.bar(x, [entries[code] for code in labels], width * n_bars, label='all entries', color=colors.pop(0))

        i = -1
        not_processed = [errors[parser_failure_label][code] * mul for code in labels]
        ax.bar(x, not_processed, width * n_bars, label=parser_failure_label, color=colors.pop(0))
        for key, values in errors.items():
            if key != parser_failure_label:
                ax.bar(x + i * width, [values[code] * mul for code in labels], width, label=key, bottom=not_processed, color=colors.pop(0))
                i += 1

        # Add some text for labels, title and custom x-axis tick labels, etc.
        ax.set_yscale('power', exponent=scale)
        ax.set_xticks(x)
        ax.set_xticklabels(labels)
        ax.legend()
        leg = ax.get_legend()
        for i in range(0, len(leg_colors)):
            leg.legendHandles[i].set_color(leg_colors[i])

        fig.tight_layout()

    ax = axs[0]
    ax.set_title('Absolute number of entries with parser errors or missing repository metadata compared to all entries per code')
    ax.set_ylabel('number of entries', )
    colors = ['grey', 'red', 'yellow', 'orange', 'brown']
    draw_error_chart(errors, ax, entries=all_entries, mul=1, scale=0.25, colors=colors)
    ax.yaxis.set_major_formatter(ticker.StrMethodFormatter('{x:,.0f}'))

    ax = axs[1]
    ax.set_title('Relative rates of entries with parser errors or missing repository metadata per code')
    ax.set_ylabel('rate in %', )
    colors = ['red', 'yellow', 'orange', 'brown']
    draw_error_chart(errors_rates, ax, mul=100, colors=colors)

    plt.show()


132
133
134
135
class Metric:
    def __init__(self, metric, label=None, power=1, multiplier=1, format=None, cumulate=False):
        if label is None:
            label = metric
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
        self.metric = metric
        self.agg = None
        self.label = label
        self.multiplier = multiplier
        self.power = power
        self.format = format
        self.cumulate = cumulate

    def draw_axis(self, axis, data, x_values, x_positions, width, color):
        value_map = {
            x: values[self.metric]
            for x, values in data[self.agg].items()
            if x in x_values}

        axis.set_yscale('power', exponent=self.power)
        axis.set_ylabel(self.label, color=color)

        if self.format is not None:
            axis.yaxis.set_major_formatter(ticker.StrMethodFormatter(self.format))

        y_values = [value_map[x] * self.multiplier for x in x_values]
        if self.cumulate:
            y_values = np.array(y_values).cumsum()
        axis.bar(x_positions, y_values, width, label=self.label, color=color)
        axis.tick_params(axis='y', labelcolor=color)


def bar_plot(client, retrieve, metric1, metric2=None, title=None):
    metrics = [] if metric1.metric == 'code_runs' else [metric1.metric]
    if metric2 is not None:
        metrics += [] if metric2.metric == 'code_runs' else [metric2.metric]

    data, x_values, agg, agg_label = retrieve(client, metrics)
    metric1.agg = agg
    if metric2 is not None:
        metric2.agg = agg
173
174

    fig, ax1 = plt.subplots(figsize=(15, 6), dpi=72)
175
176
177
178
    x = np.arange(len(x_values))
    width = 0.7 / 2
    if metric2 is None:
        width = 0.7
179
180
181
    plt.sca(ax1)
    plt.xticks(rotation=90)
    ax1.set_xticks(x)
182
183
184
185
186
187
188
    ax1.set_xticklabels(x_values)
    if title is None:
        title = 'Number of %s' % metric1.label
        if metric2 is not None:
            title += ' and %s' % metric2.label
        title += ' per %s' % agg_label
        ax1.set_title(title)
189

190
    metric1.draw_axis(ax1, data, x_values, x - (width / 2 if metric2 is not None else 0), width, 'tab:red')
191

192
193
194
    if metric2:
        ax2 = ax1.twinx()  # instantiate a second axes that shares the same x-axis
        metric2.draw_axis(ax2, data, x_values, x + width / 2, width, 'tab:blue')
195
196
197
198
199

    fig.tight_layout()
    plt.show()


200
201
@client.command(help='Generate various matplotlib charts')
@click.option('--errors', is_flag=True, help='Two charts with relative and absolute parser/normalizer errors per code.')
202
203
204
205
206
207
@click.option('--x-axis', type=str, help='Aggregation used for x-axis, values are "code" and "time".')
@click.option('--y-axis', multiple=True, type=str, help='Metrics used for y-axis, values are "entries", "energies", "users".')
@click.option('--cumulate', is_flag=True, help='Cumulate over x-axis.')
@click.option('--title', type=str, help='Override chart title with given value.')
@click.option('--total', is_flag=True, help='Provide total sums of key metrics.')
def statistics(errors, title, x_axis, y_axis, cumulate, total):
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
    from .client import create_client
    client = create_client()

    class PowerScale(mscale.ScaleBase):
        name = 'power'

        def __init__(self, axis, exponent, **kwargs):
            mscale.ScaleBase.__init__(self, axis, **kwargs)
            self.exponent = exponent

        def set_default_locators_and_formatters(self, axis):
            axis.set_major_locator(ticker.AutoLocator())
            axis.set_major_formatter(ticker.ScalarFormatter())
            axis.set_minor_locator(ticker.NullLocator())
            axis.set_minor_formatter(ticker.NullFormatter())

        def limit_range_for_scale(self, vmin, vmax, minpos):
            return max(0., vmin), vmax

        class Transform(mtransforms.Transform):
            input_dims = 1
            output_dims = 1
            is_separable = True

            def __init__(self, exponent):
                super().__init__()
                self.exponent = exponent

            def transform_non_affine(self, a):
                return np.array(a)**self.exponent

            def inverted(self):
                return PowerScale.Transform(1 / self.exponent)

        def get_transform(self):
            return self.Transform(self.exponent)

    mscale.register_scale(PowerScale)

247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
    metrics = {
        'entries': Metric(
            'code_runs',
            label='entries (code runs)',
            cumulate=cumulate,
            power=0.25 if not cumulate else 1, multiplier=1e-6, format='{x:,.1f}M'),
        'users': Metric(
            'users',
            cumulate=cumulate,
            label='users that provided data'),
        'energies': Metric(
            'total_energies',
            label='total energy calculations',
            cumulate=cumulate,
            power=0.25 if not cumulate else 1, multiplier=1e-6, format='{x:,.1f}M')
    }

264
265
    if errors:
        error_fig(client)
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283

    if x_axis is not None:
        assert 1 <= len(y_axis) <= 2, 'Need 1 or 2 y axis'

        if x_axis == 'code':
            x_axis = codes
        elif x_axis == 'time':
            x_axis = dates
        else:
            assert False, 'x axis can only be "code" or "time"'

        y_axis = [metrics[y] for y in y_axis]

        bar_plot(client, x_axis, *y_axis, title=title)

    if total:
        data = client.repo.search(per_page=1, owner='admin', metrics=['total_energies', 'users', 'datasets']).response().result
        print(data.quantities['total'])