I have tried to run new example model with the new synthetic data. Training was successful and model was created, however script for evaluation gave this error: "Usecols do not match columns, columns expected but not found: ['measurement_concept_id', 'person_id', 'value_as_number']"
Then, I have checked some files in evaluation directory but I saw that some of these files are empty. (For example observation.csv file occupies 64.6 MB on disk but it is empty).
Am I the only person who is having this error?
Thanks.
Created by egealpay @Bcragin,
You're right, I didn't catch that, thank you for bringing this to our attention! I'll updated the synthetic data with a clean version.
Thanks,
@trberg Yes, it looks to me like in the training version of the new synthetic data (but not in the evaluation version or in the UW data) the measurement.csv file has an extra, unnamed first column -- essentially a row number column. That makes it hard to write code that will run both locally and on the UW site. Hi @egealpay, I'm having a somewhat similar error when reading in the measurement.csv from the UW data. The model runs trains fine locally, reading the (new) synthetic data files (including measurement.csv) correctly. But when the the training is attempted on the UW data the run fails giving the error
Error in scan(file = file, what = what, sep = sep, quote = quote, dec = dec, :
scan() expected 'a real', got '2013-10-02'
Calls: read.csv -> read.table -> scan
Execution halted
when reading the measurement file. I have not noticed any "empty" files in the synthetic data, however.
Update: actually I need to correct part of what I said above. While the training step went fine locally, the inference part failed at the same point and with the same error as I got on the UW data. So our problems are very similar. Maybe the measurement.csv files in the training and inference subsets have slightly different formats.
Drop files to upload
New Example Model - Error running locally page is loading…