Reputation: 31
I have run through the Google Prediction API tutorials and documentation for "hello prediction - https://cloud.google.com/prediction/docs/hello_world
However when training the model in the developer console my request fails with the following output:
Request:
POST https://www.googleapis.com/prediction/v1.6/projects/959568262740/trainedmodels?key={YOUR_API_KEY}
{
"id": "language_id",
"storageDataLocation": "http://storage.googleapis.com/2341234/language_id.txt"
}
Response:
400 OK
- Show headers -
{
"error": {
"errors": [
{
"domain": "global",
"reason": "invalid",
"message": "Training data file is empty.",
"locationType": "other",
"location": "id"
}
],
"code": 400,
"message": "Training data file is empty."
}
}
I've implemented Authorize requests using OAuth 2.0: - is there anything else that I should be doing, or that may have changed between Google PRediction API v 1.6 and the tutorial. Any link to any additional tutorial or article on the subject would also be extremely valuable so can debug myself!
Upvotes: 0
Views: 351
Reputation: 121
Your storageDataLocation is not correct.
When you go to the Overview tab in Google Developers Console, you'll find your Project ID: xxxxx (example) and the bucket you've stored in is called yyyy (example).
Then replace the "storageDataLocation:" "xxxxx/yyyy". Should solve this problem. You need the relative path to Google Storage, and not the absolute web path.
Upvotes: 0