Error with 60Hz dataset continues

[KevKo1990]

Hi,

thank you @adswa for pushing version 1.1 to pip. Unfortunetaly, my error from https://github.com/psychoinformatics-de/remodnav/issues/28 persists. I checked if the new error warning occur, however, this is not the case. My variable selection causes no warning messgage.

Traceback (most recent call last):
  File ".../lib/python3.8/multiprocessing/pool.py", line 125, in worker
    result = (True, func(*args, **kwds))
  File ".../lib/python3.8/site-packages/joblib/_parallel_backends.py", line 595, in __call__
    return self.func(*args, **kwargs)
  File ".../lib/python3.8/site-packages/joblib/parallel.py", line 262, in __call__
    return [func(*args, **kwargs)
  File ".../lib/python3.8/site-packages/joblib/parallel.py", line 262, in <listcomp>
    return [func(*args, **kwargs)
  File "/local/home/kekoch/python_projects/lab_study/eye_tracking_analysis/label_eye_data/create_eye_tracking_event_data.py", line 279, in create_remodnav_event_files
    remodnav_df = apply_remodnav(tmp_read_filepath, tmp_store_filepath,
  File "/local/home/kekoch/python_projects/lab_study/eye_tracking_analysis/label_eye_data/create_eye_tracking_event_data.py", line 96, in apply_remodnav
    rem.main([sys.argv[0]] + variables_remodnav)
  File ".../lib/python3.8/site-packages/remodnav/__init__.py", line 152, in main
    events = clf(pp, classify_isp=True, sort_events=True)
  File ".../lib/python3.8/site-packages/remodnav/clf.py", line 371, in __call__
    events.extend(self._classify_intersaccade_periods(
  File ".../lib/python3.8/site-packages/remodnav/clf.py", line 554, in _classify_intersaccade_periods
    for e in self._classify_intersaccade_period(
  File ".../lib/python3.8/site-packages/remodnav/clf.py", line 596, in _classify_intersaccade_period
    for e in self._classify_intersaccade_period_helper(
  File ".../lib/python3.8/site-packages/remodnav/clf.py", line 667, in _classify_intersaccade_period_helper
    for e in self._fix_or_pursuit(data, start, end):
  File ".../lib/python3.8/site-packages/remodnav/clf.py", line 775, in _fix_or_pursuit
    yield self._mk_event_record(
  File ".../lib/python3.8/site-packages/remodnav/clf.py", line 342, in _mk_event_record
    self._get_signal_props(data[start:end])))
  File ".../lib/python3.8/site-packages/remodnav/clf.py", line 276, in _get_signal_props
    pv = data['vel'].max()
  File ".../lib/python3.8/site-packages/numpy/core/_methods.py", line 39, in _amax
    return umr_maximum(a, axis, None, out, keepdims, initial, where)
ValueError: zero-size array to reduction operation maximum which has no identity
"""

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/local/home/kekoch/python_projects/lab_study/eye_tracking_analysis/label_eye_data/create_eye_tracking_event_data.py", line 302, in <module>
    Parallel(n_jobs=NUM_CORES, backend='multiprocessing')(
  File ".../lib/python3.8/site-packages/joblib/parallel.py", line 1054, in __call__
    self.retrieve()
  File ".../lib/python3.8/site-packages/joblib/parallel.py", line 933, in retrieve
    self._output.extend(job.get(timeout=self.timeout))
  File ".../lib/python3.8/multiprocessing/pool.py", line 771, in get
    raise self._value
ValueError: zero-size array to reduction operation maximum which has no identity

I use the following parameter settings to run the algorithm (I optimized the settings based on a self-labeled dataset):

0.0181155685774209 60 --savgol-length 0.05 --median-filter-length 0.05 --velthresh-startvelocity 300 --pursuit-velthresh 5 --saccade-context-window-length 2 --dilate-nan 0.02 --min-saccade-duration 0.02

To replicate the error, I attached the crashing part of my dataset: eye_tracking.txt

Do you have any idea why with my data the algorithm crashes? Thank you already for your help!

[adswa]

Hey, thanks for the detailed issue and the example file! The issue is in your data, I think. I'm able to replicate your error, but it arises from the fact that there are empty lines in your input data. If you strip those, the classification works.

(remodnav) adina@muninn in ~/Downloads
❱ sed -i '/^\s*$/d'  eye_tracking\(1\).txt
(remodnav) adina@muninn in ~/Downloads
❱ cat eye_tracking\(1\).txt 
1054.7740745544434  567.3203587532043
1059.4937896728516  565.1391756534576
1087.8670692443848  587.2248977422714
1073.7462043762207  604.0173554420471
1066.3418197631836  582.63991355896
1139.8681640625 562.4450039863586
1140.2293395996094  563.3534359931946
1107.6242637634277  562.6706153154373
1109.509048461914   569.034161567688
1107.560634613037   569.4072031974792
1108.5669708251953  576.1746418476105
1101.1261940002441  587.4281072616577
1094.27490234375    573.6687612533569
1096.640281677246   578.154981136322
1096.1009788513184  591.5245485305786
1092.9481887817383  582.9961216449738
1092.528419494629   568.4891474246979
1083.4343147277832  577.3953473567963
1088.594913482666   579.4349098205566
1107.3037719726562  560.4691708087921
1102.074966430664   579.1347706317902
1107.9465866088867  570.4332447052002
1114.1997528076172  572.8529620170593
1099.4373321533203  585.4550743103027
1100.5931854248047  592.4616265296936
1101.7505264282227  581.5504002571106
1106.3683319091797  590.8732867240906
1126.8705368041992  604.4198155403137
1131.4482879638672  582.0984077453613
1131.4482879638672  582.0984077453613
1128.3570098876953  584.6828556060791
1127.5916290283203  579.1564965248108
1107.4720001220703  580.9134936332703
1110.9539794921875  579.1421735286713
1141.6521835327148  572.0102226734161
1151.3321685791016  577.0430660247803
1144.7561645507812  553.9843618869781
1134.9304389953613  551.4235389232635
1134.9708366394043  558.5087871551514
1134.2570686340332  562.5474536418915
1136.8513870239258  559.6298110485077
1132.6181602478027  558.2517457008362
1123.410472869873   572.329882979393
1132.7839851379395  568.1860792636871
1126.0827255249023  571.065194606781
1115.379695892334   564.9577403068542
1088.53609085083    552.6089519262314
1100.2128410339355  556.4097762107849
1122.6976203918457  560.4889011383057
1134.7531700134277  557.9530549049377
1059.6366119384766  557.4063992500305
1073.6090469360352  567.1656703948975
1091.6386985778809  601.1293888092041
1087.3554611206055  576.4352238178253
1084.1843605041504  572.0255756378174
1082.4741554260254  573.7123095989227
1058.2696723937988  576.0424518585205
1055.9429168701172  579.5147001743317
1051.8917655944824  585.9971272945404
1048.7213516235352  586.3098192214966
1039.9726867675781  592.834507226944
1033.1046867370605  593.3007287979126
1045.0513458251953  588.118861913681
1054.562759399414   578.6194324493408
1053.5430335998535  584.6184504032135
1054.3145370483398  593.9500272274017
1043.9703941345215  583.5110414028168
1044.0279579162598  582.5770854949951
1048.0718421936035  605.4385828971863
1043.2265281677246  598.452855348587
1060.1206970214844  565.0927948951721
1022.0997619628906  541.0411047935486
1027.0194625854492  585.5226016044617
1042.053508758545   596.5517270565033
1108.2175827026367  620.2361047267914
1120.1012992858887  606.204879283905
1127.1065139770508  574.5857548713684
1149.7362899780273  612.8430247306824
1115.5833435058594  620.0620079040527
1106.9770431518555  629.5080399513245
1111.227035522461   575.0091683864594
1111.2773323059082  564.1863262653351
1108.073787689209   573.1479513645172
1111.2110137939453  576.2617385387421
1109.5995712280273  568.159943819046
1108.6676788330078  564.9553263187408
1152.6319313049316  531.3502138853073
1282.9631423950195  415.2369511127472
1298.6566543579102  313.3840173482895
1272.1385192871094  238.0040037631988
1278.7775802612305  249.966521859169
1284.5343589782715  250.20884603261945
1285.8380699157715  246.68699830770493
1284.1532135009766  245.5621039867401
1295.6406784057617  227.2441506385803
1299.1432571411133  233.3820855617523
1289.5955085754395  247.2242876887321
1286.7933082580566  247.20054209232327
1291.8848419189453  249.48531746864316
1272.5455856323242  231.964253783226
1283.1925964355469  220.6830114126205
1291.179141998291   211.39079868793482
1292.134609222412   210.70489615201944
1277.74658203125    245.74874550104136
1217.1868515014648  336.2017813324928
1308.2143592834473  221.70179486274716
1308.2143592834473  221.70179486274716
1332.7217102050781  247.4624919891357
1330.345687866211   230.49596965312955
1334.5793724060059  200.69163322448728
1322.168197631836   221.88373714685437
1318.458137512207   230.0630450248718
1333.494701385498   208.80337357521051
1326.2238693237305  209.80336815118784
1329.287166595459   213.01373898983
1355.6187057495117  188.29972296953193
1355.0450134277344  218.55484753847117
1350.5336952209473  209.42072689533234
1350.8926391601562  206.39026254415504
1345.6965065002441  219.85406398773193
1346.6005325317383  210.4570841789245
1349.4575500488281  197.01427638530726
1349.5458984375 201.53403460979456
1336.1417770385742  211.07163727283475
1344.9931526184082  212.3477035760879
1340.3471374511719  219.95983690023417
1333.0910110473633  220.2659466862678
1344.2718887329102  199.6949860453605
1339.6259880065918  206.9429531693458
1335.2061080932617  232.93816924095148
1338.566722869873   208.6932152509689
1337.8807640075684  200.4255956411361
1335.8140754699707  202.33742594718927
1330.9332847595215  204.9282386898994
1328.0816459655762  217.69140422344205
1323.5829734802246  223.9583507180214
1326.936149597168   206.06628924608225
1331.8769073486328  208.5886895656585
1331.5080642700195  218.46721976995462
1342.8633499145508  219.56320255994797
1343.5133743286133  248.5138964653015
1291.6427993774414  516.9105255603789
448.0543041229248   467.4430167675018
448.0543041229248   467.4430167675018
1360.8380126953125  751.3775861263275
1360.8380126953125  751.3775861263275
1278.369369506836   756.1146676540375
1272.1778297424316  756.5988492965698
1255.6648063659668  739.158525466919
1255.2773094177246  744.6975982189178
1242.6923561096191  738.5959053039551
1240.4677391052246  746.4400148391724
1239.8784828186035  812.734147310257
1246.868019104004   758.3018374443054
1243.1351280212402  753.4046924114227
1269.7058486938477  716.8673086166382
1254.071159362793   746.2193119525909
1252.5091552734375  740.8710408210754
1252.906608581543   743.0390274524689
1256.7271614074707  744.0558314323425
(remodnav) adina@muninn in ~/Downloads
❱ remodnav eye_tracking\(1\).txt out.tsv 0.0181155685774209 60 --savgol-length 0.05 --median-filter-length 0.05 --velthresh-startvelocity 300 --pursuit-velthresh 5 --saccade-context-window-length 2 --dilate-nan 0.02 --min-saccade-duration 0.02

WARNING:Computed velocity exceeds threshold. Inappropriate filter setup? [1039.0 > 1000.0 deg/s]

Please reopen if the issue persists, and obviously feel free to open new issues if something else is wrong. :)

[KevKo1990]

Thank you for looking into it, @adswa! There are indeed empty spaces, however, the missing data is to some extent planned. In our case, we had 3 monitors in place but only the central one was recorded with a eye tracker. If people watched on the other screens (or blinked), nan (empty data points) were created.

Does Remodnav currently not handle missing data points? What is your recommendation on how to treat them?

[adswa]

Hey, thanks - you're right. The failure actually shouldn't arise. I took a closer look into your data, its a bit of an unfortunate edge case we never hit because we've not dealt with such low sampling rates. The crash seems to happen because of an unfortunate combination of a missing value followed by several identical coordinates. This causes a high jump in velocities followed by interpolation to the previous (nan) velocity value, and the median filter dealt poorly with it due to identical values that followed. In the end, an event with only nan velocity ended up in the classification, and then numpy's np.max() crashed. I pushed a basic safeguard in #36, but its not really a pretty one, because due to the nature of the data, certain properties end up not being computed for this sample (e.g., amplitude or peak velocity). I could get around this by removing one of the duplicated lines in the input data after the last nan, or by removing the last nan, but that, too, isn't really a fix. Does this happen for much of your data, or is it a rare case?

[adswa]

I have released 1.1.1 with the safeguard to PyPi.

[KevKo1990]

@adswa I tested it and it fixed the errors on my end with a 60Hz dataset. Thank you!