Comparing Workouts Missing line plot
Comparing power line plot for two workouts only displayed one line on the plot. The legend/key stated the other was NaN value. I have attached screenshots.
I checked the raw data for the one with NaN value, and the only anomaly maybe could be SPM was a negative number. Just guessing this has any impact though.
I have attached screenshots to see data involved.
Comments (37)
-
repo owner -
reporter Sure. Do you want the numbers fresh from concept 2 log or dumped back out of the Db here or both?
-
repo owner Both. Do copy support@rowsandall.com
-
reporter - attached concept2-result-43333800.csv
- attached concept2-result-43246038.csv
-
reporter raw concept2 log files added.
-
repo owner I suspect over aggressive data cleaning (note to self)
-
repo owner I am unable to reproduce this issue without knowing the exact workout number for this workout. Could you please send me the link to both workouts? Not the files, but the links on Rowsandall.com.
-
repo owner For this issue, could you also send me the direct link to the workout on rowsandall.com. thanks
-
reporter -
reporter Is these public links what you need? They are the links to the two workouts on the attached plot.
Or is there another “direct link” needed I am missing? The two workouts are just imported sessions on the concept2 , there was no workout designed inside rowsandall that these came from if that is possible?
As a side question, when you say “workout number”, what is your definition of that? Is that the name for instance “C2 Import Workout from 2020-04-07 18:10:00+00:00” or is there another unique ID|Primary key somewhere I am missing in the UI?
Thanks
Steve
-
reporter Just looking at Swagger and REST API I am guessing it is the above id =
"name": "string"
in swagger. Is that correct? -
reporter If you want my other email address for this test member, please let me know a private address I could put this too. Probably best not to have it on public tickets.
Thanks
Steve
-
repo owner Stephen, I am working on another issue right now. The workout number or ID is the code, for example a09d7609 in the example above. Sorry for being unclear. You can communicate with me on support@rowsandall.com
-
reporter No problem Sander, here to help. Thanks for the clarification on the number.
Out of interest I was starting to look at other fields of data from the concept2 and noticed ergdata isn’t storing them. Which other apps are good for capturing force data or speed etc?
I’ll do my homework to see what the alternatives are, but if you have any experience in this area as to which is good and which will waste my time, I would be grateful of any advice, thanks.
Don’t worry about not getting to issues just yet, its software support I’m in this also! (Visit Brno often as have a team and office there)…
-
repo owner That makes me wonder which company you work for … Next time you are in Brno (after the COVID stuff), give me a heads up and we’ll go for a beer (or a row, or a row and a beer). On Android, use BoatCoach. On iPhone, use PainSled. Do not pull the data through concept2.com but send them to workouts@rowsandall.com directly (from address must be the address you’re registered with). Also, check analytics.rowsandall.com for tips and tricks
-
repo owner I tried downloading the source data for both workouts and running the comparison on my development computer. That doesn’t seem to reproduce the issue
-
reporter Let me retry steps again see if I can spot anything.
-
reporter Hi
So it seems reproducible at my end.
Steps I took:
- Got to compare workout
- Select 2020-04-07 C2 Import Workout from 2020-04-07 18:10:00+00:00 4000m 00:20:47 and 2020-03-31 C2 Import Workout from 2020-03-31 17:34:00+00:00 5000m 00:29:11 in the compare workout page
- Set Yparam to Power
- Set PLot Type to Line
- Click compare
I’m looking at your plot and the distances (m) look different to the ones i clicked in the compare workout page. Not sure what is the cause there? Something to check?
The public links I put in comments seem to be the right ones, I was concerned I sent you the wrong ones before, but they also look correct.
-
repo owner @Stephen Johnson will check later. Now off to workout
-
repo owner Confirming that I am able to reproduce this behaviour (now that i have coach access I can test it)
-
repo owner I downloaded the CSV file for the 03-31 workout and uploaded it as a new workout: https://www.rowsandall.com/rowers/workout/7577ed63/
I tried to compare the 04-07 workout with the newly created workout and that worked.
I wanted to try re-importing the workout from Concept2 logbook but I couldn’t find it there.
I also went into the python shell on the server and tried to get the data. The direct loading through dataprep module worked as expected, but running the same command from within interactiveplots failed. Weird.
Do you have this issue with other workout combinations as well?
-
reporter I’ll run some checks to see if I can reproduce this with other data sets.
-
reporter The Workout type says Interval even though it was a 5000m. Could this influence? It wasn’t my session so unsure why it is interval.
-
reporter Yes I can make it happen comparing other workouts. See attached image for the two workouts..
-
reporter A number of workouts picked at random.
-
repo owner Which app are you using to record workouts? You may be better off using boatcoach or painsled and sending the data directly to rowsandall.com
Best regards, Sander
-- Met vriendelijke groet, best regards
Sander Roosendaal roosendaalsander@gmail.com
-
reporter Hi Sander
ergdata with sync into concept2.
-
reporter I reimported the same workout 2 more times as check and both show consistent non nan value. Interesting to note the newer imports are of shorter length. Could there be something in the import process do you think?
-
reporter
-
reporter It seems that any workout that has imported that has a row with ftime 00:00 (and index 0) has an issue.
-
repo owner That is a very useful observation
-
reporter I also noticed that ftime and pace for the imported Test Import 2 run i trialled was different to the original in concept2-result-43246038 attached csv file . pace for the 12m mark seemed out by 5 seconds. I’ll recheck my numbers just in case I got wrong. I assume fpace imeans the same as pace (seconds) in the csv file.
-
repo owner I just figured out that it is probably the “workoutstate” parameter that is set to 0 in those workouts. That is a Concept2 transition state. I am fixing that and deploying a fix in the next hour. After that, could you check if that fixed the issue?
-
reporter I tried reimporting from the same concept 2 log csv and the 0 row does not appear now. It looks fixed Sander.
-
repo owner Stephen, it should also be fixed for the older imports, the one you reported above. Please check and report. Thanks!
-
reporter Looks like it has resolved problem Sander. Comparing same workouts I did in a previous message they all show value now and not NaN value. Had to change to cumulative distance as one of the workouts was interval. But it looks ok Yes!
-
repo owner - changed status to resolved
Thanks for the help and patience, Stephen!
- Log in to comment
Could you send the workout numbers?