Create ML: Testing Error expected directory at URL

Question

Created Mar ’21

Replies 3

Boosts 0

Views 2.1k

Participants 4

Hi, I'm new to Create ML. I was trying to create a simple sentiment analysis model. My input file is cleaned JSON data from Apple reviews with two fields, "label" (pos/neg) and "text" (the content with sentiment). Training runs successfully. However, when I try to perform a test with similar data with the same fields in either JSON or CSV format, I continue to get the following error.

Testing Error: Expected directory at URL at "filename.json" I've tried using different data sets but always receive the same error.

Does anyone have any idea what I'm doing wrong?

Boost

Answer 1

kbulutozler OP

Aug ’21

I don't know what the error message exactly means, however i found a solution by accident. I think createML expects us to give a directory (training, validation or testing) that contains a subdirectory for each class which has the textual data in it. So, for a binary sentiment analysis problem, "train" directory should have 2 sub directories called "neg" and "pos". "neg" directory has files of examples that have negative sentiment. Likewise, "pos" directory has files of examples that have positive sentiment.

Hope this helps.

1

Answer 2

yoKurt OP

Nov ’21

Is there an answer to this problem? What directory exactly is missing? kbulutozler's answer doesn't work for me for some reason. I am using an CSV file to train a model, which worked perfectly in the past.

0

Answer 3

mueller-stephan OP

Jan ’22

I got the same error message when attempting to use the same .csv format as for training. After a little trial and error, what eventually worked was supplying the data in the following format, as e.g. used for Vision tasks in CreateML:

test_data/
├─ label_0/
│  ├─ 0.txt
│  ├─ 1.txt
│  ├─ ...
├─ label_1/
│  ├─ 0.txt
│  ├─ 1.txt
│  ├─ 2.txt
│  ├─ ...

Here, the files 0.txt, ... containing only the text strings. Hope this helps you!

1