I have a project about time series analysis. My data are not stationary and they have daily seasonality as shown in figure below. Is it correct to do the following steps?
- Decompose Time serie into season, trend , residuals
- Model residuals using ARIMA
- Then add back seasonality and trend
Should i model seasonality and trends too?
Count
12
18
7
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
11
25
39
63
163
190
219
289
416
600
679
705
770
807
720
544
603
592
419
333
336
424
290
369
273
342
269
297
274
331
291
357
185
301
287
399
342
338
293
452
331
545
390
355
362
385
443
449
400
594
405
551
355
407
324
391
206
307
201
223
213
209
194
163
166
156
119
126
91
71
9
37
45
23
33
9
27
3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
14
18
45
60
138
157
190
272
395
556
516
553
811
702
630
507
480
425
349
309
310
379
282
248
256
248
226
222
205
242
214
220
236
166
276
337
239
309
371
290
397
299
343
377
358
436
386
346
467
571
424
398
304
346
323
225
264
258
171
168
109
200
130
94
93
94
139
87
56
37
77
25
42
26
32
16
25
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
15
10
40
61
126
172
208
254
345
568
567
634
616
808
728
398
455
440
350
320
289
386
259
290
249
344
226
296
235
279
219
284
171
240
225
350
316
335
311
301
350
356
390
308
366
443
285
396
416
468
466
401
292
334
454
309
296
294
214
235
167
248
144
141
124
110
222
108
88
87
31
31
38
40
41