Closed komatded closed 2 years ago
Hi @komatded, what you see in the URL for a prediction is what we call the prediction's id
, which is unique and automatically generated. It is intentionally different from the identifier
for that prediction. The id
is always unique, but it's possible, though discouraged, for you to log multiple predictions with the same identifier
. Does that make sense?
I just tried the example in languages/python/examples/advanced
which logs a single prediction and true value to the app running locally, and was able to successfully associate the true value with the prediction using the identifier
. Perhaps there is something different in your setup? Can you provide some more detailed steps to reproduce your problem? Thank you! :)
Hi @nitsky, thank you for your response!
I apologize for disturbing in vain :) Looks like I found what the problem was - in my train and test datasets target's column data type was integer
, so I thought that I should also pass integer as true_value
in log_true_value
... now when I've passed it as a string everything worked fine!
Sorry, for bothering! I guess now this issue can be closed
@komatded if the value is an integer you should not be required to pass it as a string. Is your model doing regression or classification?
Classification; I've trained it with command tangram train --file web_train_tangram.csv --target target --output web.tangram
, and as you can see target columns data type is int64
Good evening!
I recently tried to push model predictions into an application and discovered that identifiers I specified in
log_prediction
function did not match identifiers that appeared in the application. I usepython3.9
andtangram==0.7.0
.Here are screenshots: