I am trying to use the Google Prediction API. I have trained my model, and tested a prediction through the web page and it works great. However, I am now trying to use the java api to predict a bunch of records but I keep getting an error
com.google.api.client.googleapis.json.GoogleJsonResponseException: 400 Bad Request
{
"code" : 400,
"errors" : [ {
"domain" : "global",
"message" : "Invalid value for: Unable to parse '[feature1, feature2, feature3, feature4, feature5]'.",
"reason" : "invalid"
} ],
"message" : "Invalid value for: Unable to parse '[feature1, feature2, feature3, feature4, feature5]'."
To me it seems like the json creator is not putting quotes around the features, but I am following the sample as close as possible and they don't change or modify the json factory. Here is the credential and prediction building code.
private static GoogleCredential authorize() throws Exception {
GoogleCredential credential = new GoogleCredential.Builder().setTransport(httpTransport)
.setJsonFactory(JSON_FACTORY)
.setServiceAccountId(SERVICE_ACCOUNT_EMAIL)
.setServiceAccountScopes(Collections.singleton(PredictionScopes.PREDICTION))
.setServiceAccountPrivateKeyFromP12File(new File("p12filefromdevconsole.p12"))
.build();
return credential;
}
...
Prediction prediction = new Prediction.Builder(
httpTransport, JSON_FACTORY, credential).setApplicationName(APPLICATION_NAME).build();
...
private static Output predict(Prediction prediction, String... features) throws IOException {
Input input = new Input();
InputInput inputInput = new InputInput();
inputInput.setCsvInstance(Collections.<Object>singletonList(features));
input.setInput(inputInput);
Output output = prediction.trainedmodels().predict(PROJECT_ID, MODEL_ID, input).execute();
return output;
}
Any thoughts what I am doing wrong?