pytrendy 2.0.0.dev3__tar.gz → 2.0.0.dev5__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/PKG-INFO +16 -16
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/README.md +15 -15
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pyproject.toml +1 -1
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/io/plot_pytrendy.py +45 -19
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/io/results_pytrendy.py +17 -26
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/post_processing/segments_analyse.py +7 -7
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/post_processing/segments_refine/segment_grouping.py +3 -7
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/LICENSE +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/__init__.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/detect_trends.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/io/__init__.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/io/data/classes_signals.csv +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/io/data/series_synthetic.csv +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/io/data_loader.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/io/prepare_index.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/post_processing/__init__.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/post_processing/segments_get.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/post_processing/segments_refine/__init__.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/post_processing/segments_refine/abrupt_shaving.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/post_processing/segments_refine/artifact_cleanup.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/post_processing/segments_refine/gradual_expand_contract.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/post_processing/segments_refine/trend_classify.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/post_processing/segments_refine/update_neighbours.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/process_signals.py +0 -0
- {pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/simpledtw.py +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: pytrendy
|
|
3
|
-
Version: 2.0.0.
|
|
3
|
+
Version: 2.0.0.dev5
|
|
4
4
|
Summary: Trend Detection in Python. Applicable for real-world industry use cases in time series.
|
|
5
5
|
License: MIT License
|
|
6
6
|
|
|
@@ -126,16 +126,16 @@ The best detected trend is Down between dates 2025-05-09 - 2025-06-17
|
|
|
126
126
|
|
|
127
127
|
Full Results:
|
|
128
128
|
-------------------------------------------------------------------------------
|
|
129
|
-
direction start end
|
|
130
|
-
time_index
|
|
131
|
-
1 Up 2025-01-02 2025-01-24
|
|
132
|
-
2 Down 2025-01-25 2025-02-05
|
|
133
|
-
3 Flat 2025-02-06 2025-02-09
|
|
134
|
-
4 Up 2025-02-10 2025-03-17
|
|
135
|
-
5 Down 2025-03-18 2025-04-01
|
|
136
|
-
6 Up 2025-04-02 2025-05-08
|
|
137
|
-
7 Down 2025-05-09 2025-06-17
|
|
138
|
-
8 Flat 2025-06-18 2025-06-30
|
|
129
|
+
direction start end steps total_change change_rank trend_class
|
|
130
|
+
time_index
|
|
131
|
+
1 Up 2025-01-02 2025-01-24 22 14.013348 5 gradual
|
|
132
|
+
2 Down 2025-01-25 2025-02-05 11 -13.564214 6 gradual
|
|
133
|
+
3 Flat 2025-02-06 2025-02-09 3 -1.168831 8 NaN
|
|
134
|
+
4 Up 2025-02-10 2025-03-17 35 26.190476 3 gradual
|
|
135
|
+
5 Down 2025-03-18 2025-04-01 14 -22.721861 4 gradual
|
|
136
|
+
6 Up 2025-04-02 2025-05-08 36 72.611833 2 gradual
|
|
137
|
+
7 Down 2025-05-09 2025-06-17 39 -73.253968 1 gradual
|
|
138
|
+
8 Flat 2025-06-18 2025-06-30 12 3.910534 7 NaN
|
|
139
139
|
-------------------------------------------------------------------------------
|
|
140
140
|
```
|
|
141
141
|
|
|
@@ -144,11 +144,11 @@ Explore the strongest uptrends:
|
|
|
144
144
|
results.filter_segments(direction='Up', sort_by='change_rank')[:3]
|
|
145
145
|
```
|
|
146
146
|
|
|
147
|
-
| time_index | direction | start | end | trend_class |
|
|
148
|
-
|
|
149
|
-
| 7 | Up | 2025-04-02 | 2025-05-08 | gradual |
|
|
150
|
-
| 4 | Up | 2025-02-10 | 2025-03-17 | gradual |
|
|
151
|
-
| 1 | Up | 2025-01-02 | 2025-01-24 | gradual |
|
|
147
|
+
| time_index | direction | start | end | trend_class | pct_change | steps | total_change | SNR | change_rank |
|
|
148
|
+
|---|---|---|---|---|---|---|---|---|---|
|
|
149
|
+
| 7 | Up | 2025-04-02 | 2025-05-08 | gradual | 367.50% | 36 | 72.61 | 21.70 | 2 |
|
|
150
|
+
| 4 | Up | 2025-02-10 | 2025-03-17 | gradual | 179.93% | 35 | 26.19 | 18.61 | 3 |
|
|
151
|
+
| 1 | Up | 2025-01-02 | 2025-01-24 | gradual | 104.41% | 22 | 14.01 | 22.21 | 5 |
|
|
152
152
|
|
|
153
153
|
`filter_segments` ranks segments by magnitude (`change_rank`). See the [API reference](https://russellsb.github.io/pytrendy/main/reference/pytrendy/io/results_pytrendy/#pytrendy.io.results_pytrendy.PyTrendyResults.filter_segments) for all filter and sort options.
|
|
154
154
|
|
|
@@ -62,16 +62,16 @@ The best detected trend is Down between dates 2025-05-09 - 2025-06-17
|
|
|
62
62
|
|
|
63
63
|
Full Results:
|
|
64
64
|
-------------------------------------------------------------------------------
|
|
65
|
-
direction start end
|
|
66
|
-
time_index
|
|
67
|
-
1 Up 2025-01-02 2025-01-24
|
|
68
|
-
2 Down 2025-01-25 2025-02-05
|
|
69
|
-
3 Flat 2025-02-06 2025-02-09
|
|
70
|
-
4 Up 2025-02-10 2025-03-17
|
|
71
|
-
5 Down 2025-03-18 2025-04-01
|
|
72
|
-
6 Up 2025-04-02 2025-05-08
|
|
73
|
-
7 Down 2025-05-09 2025-06-17
|
|
74
|
-
8 Flat 2025-06-18 2025-06-30
|
|
65
|
+
direction start end steps total_change change_rank trend_class
|
|
66
|
+
time_index
|
|
67
|
+
1 Up 2025-01-02 2025-01-24 22 14.013348 5 gradual
|
|
68
|
+
2 Down 2025-01-25 2025-02-05 11 -13.564214 6 gradual
|
|
69
|
+
3 Flat 2025-02-06 2025-02-09 3 -1.168831 8 NaN
|
|
70
|
+
4 Up 2025-02-10 2025-03-17 35 26.190476 3 gradual
|
|
71
|
+
5 Down 2025-03-18 2025-04-01 14 -22.721861 4 gradual
|
|
72
|
+
6 Up 2025-04-02 2025-05-08 36 72.611833 2 gradual
|
|
73
|
+
7 Down 2025-05-09 2025-06-17 39 -73.253968 1 gradual
|
|
74
|
+
8 Flat 2025-06-18 2025-06-30 12 3.910534 7 NaN
|
|
75
75
|
-------------------------------------------------------------------------------
|
|
76
76
|
```
|
|
77
77
|
|
|
@@ -80,11 +80,11 @@ Explore the strongest uptrends:
|
|
|
80
80
|
results.filter_segments(direction='Up', sort_by='change_rank')[:3]
|
|
81
81
|
```
|
|
82
82
|
|
|
83
|
-
| time_index | direction | start | end | trend_class |
|
|
84
|
-
|
|
85
|
-
| 7 | Up | 2025-04-02 | 2025-05-08 | gradual |
|
|
86
|
-
| 4 | Up | 2025-02-10 | 2025-03-17 | gradual |
|
|
87
|
-
| 1 | Up | 2025-01-02 | 2025-01-24 | gradual |
|
|
83
|
+
| time_index | direction | start | end | trend_class | pct_change | steps | total_change | SNR | change_rank |
|
|
84
|
+
|---|---|---|---|---|---|---|---|---|---|
|
|
85
|
+
| 7 | Up | 2025-04-02 | 2025-05-08 | gradual | 367.50% | 36 | 72.61 | 21.70 | 2 |
|
|
86
|
+
| 4 | Up | 2025-02-10 | 2025-03-17 | gradual | 179.93% | 35 | 26.19 | 18.61 | 3 |
|
|
87
|
+
| 1 | Up | 2025-01-02 | 2025-01-24 | gradual | 104.41% | 22 | 14.01 | 22.21 | 5 |
|
|
88
88
|
|
|
89
89
|
`filter_segments` ranks segments by magnitude (`change_rank`). See the [API reference](https://russellsb.github.io/pytrendy/main/reference/pytrendy/io/results_pytrendy/#pytrendy.io.results_pytrendy.PyTrendyResults.filter_segments) for all filter and sort options.
|
|
90
90
|
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
[project]
|
|
2
2
|
name = "pytrendy"
|
|
3
|
-
version = "2.0.0.
|
|
3
|
+
version = "2.0.0.dev5"
|
|
4
4
|
description = "Trend Detection in Python. Applicable for real-world industry use cases in time series."
|
|
5
5
|
authors = [
|
|
6
6
|
{ name = "Russell Sammut Bonnici", email = "r.sammutbonnici@gmail.com" },
|
|
@@ -119,9 +119,13 @@ def plot_pytrendy(df: pd.DataFrame, value_col: str, segments_enhanced: list[dict
|
|
|
119
119
|
|
|
120
120
|
# Add shaded regions with fill_between
|
|
121
121
|
ymin, ymax = ax.get_ylim() # get plot's visible y-range
|
|
122
|
+
# 'date' and 'datetime64' are both real date axes: boundaries must be
|
|
123
|
+
# displaced by the actual point spacing (via _adjacent_to), not a hard-coded
|
|
124
|
+
# one-day step, or non-daily data leaves white gaps between shaded regions.
|
|
125
|
+
is_date_axis = index_type in ('date', 'datetime64')
|
|
122
126
|
for i, seg in enumerate(segments_enhanced):
|
|
123
127
|
|
|
124
|
-
if
|
|
128
|
+
if is_date_axis:
|
|
125
129
|
start = pd.to_datetime(seg['start'])
|
|
126
130
|
end = pd.to_datetime(seg['end'])
|
|
127
131
|
else:
|
|
@@ -132,13 +136,14 @@ def plot_pytrendy(df: pd.DataFrame, value_col: str, segments_enhanced: list[dict
|
|
|
132
136
|
|
|
133
137
|
# Get context on prev seg if possible
|
|
134
138
|
prev_seg = segments_enhanced[i-1] if i-1 >= 0 else None
|
|
135
|
-
if
|
|
139
|
+
if is_date_axis:
|
|
136
140
|
prev_point = _adjacent_to(df.index, start, -1)
|
|
137
141
|
prev_neighbouring = prev_seg and (prev_point is not None) and (pd.to_datetime(prev_seg['end']) == prev_point)
|
|
138
142
|
elif index_type == 'string':
|
|
139
143
|
prev_neighbouring = prev_seg and (prev_seg['end'] == df.index[df.index.get_loc(start) - 1])
|
|
140
144
|
else:
|
|
141
|
-
|
|
145
|
+
prev_point = _adjacent_to(df.index, start, -1)
|
|
146
|
+
prev_neighbouring = prev_seg and (prev_point is not None) and (prev_seg['end'] == prev_point)
|
|
142
147
|
|
|
143
148
|
is_prev_not_trend = prev_seg and (not ('trend_class' in prev_seg))
|
|
144
149
|
|
|
@@ -149,14 +154,15 @@ def plot_pytrendy(df: pd.DataFrame, value_col: str, segments_enhanced: list[dict
|
|
|
149
154
|
|
|
150
155
|
# Get context on next seg if possible
|
|
151
156
|
next_seg = segments_enhanced[i+1] if i+1 < len(segments_enhanced) else None
|
|
152
|
-
if
|
|
157
|
+
if is_date_axis:
|
|
153
158
|
next_point = _adjacent_to(df.index, end, 1)
|
|
154
159
|
next_neighbouring = next_seg and (next_point is not None) and (pd.to_datetime(next_seg['start']) == next_point)
|
|
155
160
|
elif index_type == 'string':
|
|
156
161
|
end_pos = df.index.get_loc(end)
|
|
157
162
|
next_neighbouring = next_seg and (next_seg['start'] == _safe_adjacent(df.index, end_pos, 1))
|
|
158
163
|
else:
|
|
159
|
-
|
|
164
|
+
next_point = _adjacent_to(df.index, end, 1)
|
|
165
|
+
next_neighbouring = next_seg and (next_point is not None) and (next_seg['start'] == next_point)
|
|
160
166
|
|
|
161
167
|
next_seg_abrupt = next_seg and (('trend_class' in next_seg) and (next_seg['trend_class'] == 'abrupt'))
|
|
162
168
|
next_seg_noise = next_seg and (next_seg['direction'] == 'Noise')
|
|
@@ -165,13 +171,13 @@ def plot_pytrendy(df: pd.DataFrame, value_col: str, segments_enhanced: list[dict
|
|
|
165
171
|
if is_abrupt or is_noise:
|
|
166
172
|
pass # Keep start as-is for abrupt/noise segments
|
|
167
173
|
else:
|
|
168
|
-
if
|
|
174
|
+
if is_date_axis:
|
|
169
175
|
new_start = _adjacent_to(df.index, start, -1) # Everything else displaced left start
|
|
170
176
|
elif index_type == 'string':
|
|
171
177
|
start_pos = df.index.get_loc(start)
|
|
172
178
|
new_start = _safe_adjacent(df.index, start_pos, -1)
|
|
173
179
|
else:
|
|
174
|
-
new_start = start -
|
|
180
|
+
new_start = _adjacent_to(df.index, start, -1) # Everything else displaced left start
|
|
175
181
|
|
|
176
182
|
# Check validity of plot start adjustment
|
|
177
183
|
value_new_start = df.loc[new_start, value_col] if new_start is not None and new_start in df.index else None
|
|
@@ -185,7 +191,7 @@ def plot_pytrendy(df: pd.DataFrame, value_col: str, segments_enhanced: list[dict
|
|
|
185
191
|
else:
|
|
186
192
|
# if not displaced and prev is not trend, adjust by plotting (as prev has already been drawn)
|
|
187
193
|
if is_prev_not_trend and prev_neighbouring:
|
|
188
|
-
if
|
|
194
|
+
if is_date_axis:
|
|
189
195
|
prev_end = pd.to_datetime(segments_enhanced[i-1]['end'])
|
|
190
196
|
prev_adj_end = _adjacent_to(df.index, prev_end, 1)
|
|
191
197
|
prev_new_end = prev_adj_end.strftime('%Y-%m-%d') if prev_adj_end is not None else None
|
|
@@ -195,7 +201,7 @@ def plot_pytrendy(df: pd.DataFrame, value_col: str, segments_enhanced: list[dict
|
|
|
195
201
|
prev_new_end = _safe_adjacent(df.index, prev_end_pos, 1)
|
|
196
202
|
else:
|
|
197
203
|
prev_end = segments_enhanced[i-1]['end']
|
|
198
|
-
prev_new_end = prev_end
|
|
204
|
+
prev_new_end = _adjacent_to(df.index, prev_end, 1)
|
|
199
205
|
|
|
200
206
|
if prev_new_end is not None:
|
|
201
207
|
if index_type == 'string':
|
|
@@ -207,13 +213,13 @@ def plot_pytrendy(df: pd.DataFrame, value_col: str, segments_enhanced: list[dict
|
|
|
207
213
|
|
|
208
214
|
# Adjust ends when appropriate
|
|
209
215
|
if (next_seg_abrupt or next_seg_noise) and next_neighbouring:
|
|
210
|
-
if
|
|
216
|
+
if is_date_axis:
|
|
211
217
|
new_end = _adjacent_to(df.index, end, 1)
|
|
212
218
|
elif index_type == 'string':
|
|
213
219
|
end_pos = df.index.get_loc(end)
|
|
214
220
|
new_end = _safe_adjacent(df.index, end_pos, 1)
|
|
215
221
|
else:
|
|
216
|
-
new_end = end
|
|
222
|
+
new_end = _adjacent_to(df.index, end, 1)
|
|
217
223
|
|
|
218
224
|
# Check validity of plot end adjustment
|
|
219
225
|
value_new_end = df.loc[new_end, value_col] if new_end is not None and new_end in df.index else None
|
|
@@ -227,7 +233,7 @@ def plot_pytrendy(df: pd.DataFrame, value_col: str, segments_enhanced: list[dict
|
|
|
227
233
|
else:
|
|
228
234
|
# if not displaced and next is noise, adjust for next plotting round
|
|
229
235
|
if next_seg_noise and next_neighbouring:
|
|
230
|
-
if
|
|
236
|
+
if is_date_axis:
|
|
231
237
|
next_start = pd.to_datetime(segments_enhanced[i+1]['start'])
|
|
232
238
|
next_adj_start = _adjacent_to(df.index, next_start, -1)
|
|
233
239
|
if next_adj_start is not None:
|
|
@@ -236,7 +242,10 @@ def plot_pytrendy(df: pd.DataFrame, value_col: str, segments_enhanced: list[dict
|
|
|
236
242
|
next_start_pos = df.index.get_loc(segments_enhanced[i+1]['start'])
|
|
237
243
|
segments_enhanced[i+1]['start'] = _safe_adjacent(df.index, next_start_pos, -1)
|
|
238
244
|
else:
|
|
239
|
-
|
|
245
|
+
next_start = segments_enhanced[i+1]['start']
|
|
246
|
+
next_adj_start = _adjacent_to(df.index, next_start, -1)
|
|
247
|
+
if next_adj_start is not None:
|
|
248
|
+
segments_enhanced[i+1]['start'] = next_adj_start
|
|
240
249
|
else:
|
|
241
250
|
pass # Keep end as-is
|
|
242
251
|
|
|
@@ -265,7 +274,7 @@ def plot_pytrendy(df: pd.DataFrame, value_col: str, segments_enhanced: list[dict
|
|
|
265
274
|
|
|
266
275
|
# Add vertical line if next seg is same & touching
|
|
267
276
|
if next_seg and next_neighbouring and next_seg['direction'] == seg['direction']:
|
|
268
|
-
if
|
|
277
|
+
if is_date_axis:
|
|
269
278
|
line_date = pd.to_datetime(seg['end'])
|
|
270
279
|
else:
|
|
271
280
|
line_date = seg['end']
|
|
@@ -282,13 +291,30 @@ def plot_pytrendy(df: pd.DataFrame, value_col: str, segments_enhanced: list[dict
|
|
|
282
291
|
ax.set_xlim(first_date, last_date)
|
|
283
292
|
ax.set_ylim(ymin, ymax)
|
|
284
293
|
|
|
285
|
-
if index_type
|
|
286
|
-
|
|
287
|
-
|
|
294
|
+
if index_type in ('date', 'datetime64'):
|
|
295
|
+
index = df.index
|
|
296
|
+
gap_days = (
|
|
297
|
+
np.median(np.diff(index.values)) / np.timedelta64(1, 'D')
|
|
298
|
+
if len(index) > 1 else 1
|
|
299
|
+
)
|
|
300
|
+
|
|
301
|
+
if gap_days <= 1:
|
|
302
|
+
# Daily data: weekly majors on the default weekday, as before.
|
|
303
|
+
ax.xaxis.set_major_locator(mdates.WeekdayLocator(interval=1))
|
|
304
|
+
else:
|
|
305
|
+
# Non-daily spacing: pin majors to the data's own index positions so
|
|
306
|
+
# every major (and its 'major' gridline) lands exactly on an
|
|
307
|
+
# observation, for any cadence including irregular. A WeekdayLocator
|
|
308
|
+
# anchors its interval grid to the Unix epoch, which lands majors
|
|
309
|
+
# beside fortnightly / month-end observations.
|
|
310
|
+
ax.set_xticks(index)
|
|
288
311
|
ax.xaxis.set_major_formatter(mdates.DateFormatter('%Y-%m-%d'))
|
|
289
312
|
|
|
290
|
-
# Minor ticks: every day
|
|
291
|
-
|
|
313
|
+
# Minor ticks: every day, but only when points are daily-or-finer. A daily
|
|
314
|
+
# ruler under non-daily data (e.g. weekly points) misrepresents the sampling,
|
|
315
|
+
# so non-daily spacing gets no minors at all.
|
|
316
|
+
if gap_days <= 1:
|
|
317
|
+
ax.xaxis.set_minor_locator(mdates.DayLocator())
|
|
292
318
|
|
|
293
319
|
# Rotate major tick labels
|
|
294
320
|
plt.setp(ax.get_xticklabels(), rotation=90, ha='right')
|
|
@@ -23,8 +23,8 @@ class PyTrendyResults:
|
|
|
23
23
|
List of dictionaries representing individual trend segments.
|
|
24
24
|
index_type (str):
|
|
25
25
|
The type of the index used for the segments (``'date'``, ``'datetime64'``,
|
|
26
|
-
``'integer'``, ``'float'``, or ``'string'``). Used to
|
|
27
|
-
|
|
26
|
+
``'integer'``, ``'float'``, or ``'string'``). Used to label segment boundaries
|
|
27
|
+
in ``print_summary`` (e.g. ``'dates'`` vs ``'indexes'``). Defaults to ``'date'``.
|
|
28
28
|
"""
|
|
29
29
|
self.segments = segments
|
|
30
30
|
self.trend_segments = [seg for seg in self.segments if 'trend_class' in seg] # Get segments that are trends (exclude flats and noise)
|
|
@@ -37,11 +37,11 @@ class PyTrendyResults:
|
|
|
37
37
|
|
|
38
38
|
def set_best(self) -> None:
|
|
39
39
|
"""
|
|
40
|
-
|
|
41
|
-
|
|
42
|
-
|
|
43
|
-
|
|
44
|
-
|
|
40
|
+
Selects the best trend segment by lowest ``change_rank``.
|
|
41
|
+
|
|
42
|
+
Sets ``self.best`` to the trend segment with the lowest ``change_rank`` (derived from
|
|
43
|
+
each segment's cumulative ``total_change``). Sets ``self.best`` to ``None`` when no
|
|
44
|
+
trend segments exist.
|
|
45
45
|
"""
|
|
46
46
|
if len(self.trend_segments) == 0:
|
|
47
47
|
self.best = None
|
|
@@ -50,10 +50,11 @@ class PyTrendyResults:
|
|
|
50
50
|
|
|
51
51
|
def set_summary(self) -> None:
|
|
52
52
|
"""
|
|
53
|
-
Computes and stores summary statistics for trend segments
|
|
53
|
+
Computes and stores summary statistics for trend segments.
|
|
54
54
|
|
|
55
|
-
|
|
56
|
-
|
|
55
|
+
Sets ``self.summary`` to counts by direction and trend class, plus the highest total
|
|
56
|
+
change. Sets ``self.df_summary`` to a tabular overview of the segments, indexed by
|
|
57
|
+
``time_index``.
|
|
57
58
|
"""
|
|
58
59
|
summary = {}
|
|
59
60
|
|
|
@@ -77,16 +78,7 @@ class PyTrendyResults:
|
|
|
77
78
|
# Set summary df (without extra details)
|
|
78
79
|
df = pd.DataFrame(self.segments)
|
|
79
80
|
|
|
80
|
-
|
|
81
|
-
'date': 'days',
|
|
82
|
-
'integer': 'index steps',
|
|
83
|
-
'float': 'index steps',
|
|
84
|
-
'string': 'index steps',
|
|
85
|
-
}.get(self.index_type, 'days')
|
|
86
|
-
|
|
87
|
-
df = df.rename({'days' : unit_descriptor}, axis = 1)
|
|
88
|
-
|
|
89
|
-
cols = ['time_index', 'direction', 'start', 'end', unit_descriptor, 'total_change', 'change_rank']
|
|
81
|
+
cols = ['time_index', 'direction', 'start', 'end', 'steps', 'total_change', 'change_rank']
|
|
90
82
|
if len(changes) > 1: # only include trend_class if atleast one trend exists
|
|
91
83
|
cols += ['trend_class']
|
|
92
84
|
df = df[cols]
|
|
@@ -129,11 +121,11 @@ class PyTrendyResults:
|
|
|
129
121
|
|
|
130
122
|
def set_df(self) -> pd.DataFrame | None:
|
|
131
123
|
"""
|
|
132
|
-
Converts
|
|
124
|
+
Converts the segment list into a pandas ``DataFrame`` for downstream analysis and export.
|
|
133
125
|
|
|
134
|
-
|
|
135
|
-
|
|
136
|
-
|
|
126
|
+
Returns:
|
|
127
|
+
pd.DataFrame | None: An empty ``DataFrame`` when there are no segments; otherwise
|
|
128
|
+
``None``, with ``self.df`` set as a side effect.
|
|
137
129
|
"""
|
|
138
130
|
# Exit if nothing to report on
|
|
139
131
|
if len(self.segments) == 0:
|
|
@@ -156,8 +148,7 @@ class PyTrendyResults:
|
|
|
156
148
|
Output format. `'df'` returns a DataFrame, `'dict'` returns a list of dictionaries.
|
|
157
149
|
|
|
158
150
|
Returns:
|
|
159
|
-
|
|
160
|
-
|
|
151
|
+
list | pd.DataFrame: Filtered and sorted segments in the specified format.
|
|
161
152
|
"""
|
|
162
153
|
segments = self.segments
|
|
163
154
|
if len(segments) == 0:
|
|
@@ -19,7 +19,7 @@ def analyse_segments(df: pd.DataFrame, value_col: str, segments: list[dict]) ->
|
|
|
19
19
|
|
|
20
20
|
- Percent change (based on start/end values)
|
|
21
21
|
|
|
22
|
-
- Duration in
|
|
22
|
+
- Duration in index steps
|
|
23
23
|
|
|
24
24
|
- Cumulative total change (sum of diffs)
|
|
25
25
|
|
|
@@ -40,7 +40,7 @@ def analyse_segments(df: pd.DataFrame, value_col: str, segments: list[dict]) ->
|
|
|
40
40
|
Returns:
|
|
41
41
|
list:
|
|
42
42
|
A list of enhanced segment dictionaries with additional keys:
|
|
43
|
-
- `'total_change'`, `'pct_change'`, `'mltp_change'`, `'
|
|
43
|
+
- `'total_change'`, `'pct_change'`, `'mltp_change'`, `'steps'`, `'SNR'`, `'change_rank'`
|
|
44
44
|
"""
|
|
45
45
|
segments_enhanced = []
|
|
46
46
|
for segment in segments:
|
|
@@ -53,11 +53,11 @@ def analyse_segments(df: pd.DataFrame, value_col: str, segments: list[dict]) ->
|
|
|
53
53
|
segment_enhanced['pct_change'] = (float(val_end / val_start - 1) if val_start != 0 else np.nan)
|
|
54
54
|
segment_enhanced['mltp_change'] = (float(val_end / val_start) if val_start != 0 else np.nan)
|
|
55
55
|
|
|
56
|
-
# Calculate
|
|
57
|
-
|
|
58
|
-
if
|
|
59
|
-
|
|
60
|
-
segment_enhanced['
|
|
56
|
+
# Calculate steps & cumulative total change
|
|
57
|
+
steps = segment['end'] - segment['start']
|
|
58
|
+
if steps == 0:
|
|
59
|
+
steps = 1 # edge case for 1 step flat between noise spike & trend
|
|
60
|
+
segment_enhanced['steps'] = steps # set steps
|
|
61
61
|
|
|
62
62
|
# Calculate cumulative total change
|
|
63
63
|
segment_enhanced['total_change'] = float(df_segment[value_col].diff().sum())
|
|
@@ -12,13 +12,9 @@ def group_segments(segments: list[dict]) -> list[dict]:
|
|
|
12
12
|
"""
|
|
13
13
|
Groups consecutive segments with the same direction if their gap is small.
|
|
14
14
|
|
|
15
|
-
Segments are grouped
|
|
16
|
-
|
|
17
|
-
|
|
18
|
-
- Their gap is ≤ `GROUPING_DISTANCE`
|
|
19
|
-
- They are not classified as `'abrupt'`
|
|
20
|
-
|
|
21
|
-
This reduces fragmentation caused by short, noisy segments.
|
|
15
|
+
Segments are grouped when they share the same direction, their gap is ≤
|
|
16
|
+
``GROUPING_DISTANCE``, and neither is classified as abrupt. This reduces fragmentation
|
|
17
|
+
caused by short, noisy segments.
|
|
22
18
|
|
|
23
19
|
Args:
|
|
24
20
|
segments (list): List of segment dictionaries.
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
{pytrendy-2.0.0.dev3 → pytrendy-2.0.0.dev5}/pytrendy/post_processing/segments_refine/__init__.py
RENAMED
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|