Azure-docs: Need guidance for misalignment with DataFrameFrequencyException

[swanderz]

I was able to figure this out, but I imagine other users would struggle to solve this exception:
azureml.automl.core.shared.forecasting_exception.DataFrameFrequencyException

"In grain(s) ('groupby1', 'groupby2'), one or more date(s) do not align with the desired frequency <8 * MonthEnds>"

@CESARDELATORRE

---

Document Details

⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

• ID: 6275c379-c934-8a50-9954-eba4527db79c
• Version Independent ID: 6a105809-571c-6c59-7f9d-4ee039adf2a2
• Content: Troubleshoot module errors - Azure Machine Learning
• Content Source: articles/machine-learning/algorithm-module-reference/designer-error-codes.md
• Service: machine-learning
• Sub-service: core
• GitHub Login: @likebupt
• Microsoft Alias: keli19

[PeterCLu]

Hi @swanderz, this is filed under the designer error codes. Did you see this exception while using the drag and drop designer?

[swanderz]

I saw it after using the drag and drop designer. If the error was listed on this page, it wouldn't have filed this issue.

[maarten2015]

I'm facing more or less the same issue; and have no clue what the supported frequencies are?
I've tried second and millisecond precision in the dataset, but without success. So I'm wondering what the supported frequencies actually are?

azureml.automl.core.shared.forecasting_exception.DataFrameFrequencyException: DataFrameFrequencyException:
    Message: The frequency is not supported.
    InnerException: None
    ErrorResponse 
{
    "error": {
        "code": "UserError",
        "inner_error": {
            "code": "InvalidData",
            "inner_error": {
                "code": "TimeFrequencyCanNotBeInferrable"
            }
        },
        "message": "The frequency is not supported."
    }
}

[PeterCLu]

Hi @swanderz and @maarten2015. Automated ML is part of the studio, but it is not part of the designer. The designer is a visual way to author ML pipelines and has a separate errors from Automated ML.

We will look into your issues.

assign:@nibaccam

[nibaccam]

@cartacioS @Aniththa looks like the error being thrown is related to autoML forecasting. Can you help here?

[Aniththa]

@swanderz @maarten2015 The team is currently looking into this. Are you using transactional data?

[swanderz]

@Aniththa my data wasn't transactional as far as I'm concerned...

[cartacioS]

Hi @maarten2015 , so any frequency (i.e. second, hour, day, etc) is fine for AutoML. That's not a worry here, but instead it's the consistency of your frequency that AutoML isn't able to process.

The original error in this post saying that while you may have some frequency for a majority of your grains, the two grains "grouby1 and grouby2" don't follow the same frequency close enough to the rest of your data. 90% of each time-series must have a frequency consistent.

Now your exact error doesn't complain about grains, likely because you didn't specify any. But the issue is the same, while your data may not be exactly transactional, we have strict frequency requirements. Ensure your frequency is consistent (every 10 seconds on the dot, every day- not skipping, etc).

[nibaccam]

@swanderz and @maarten2015 thank you for reporting this issue. A bug has been filed with our development team and I'll work with them to get this detail documented. We will now proceed to close this thread. #please-close

[putriafebriana98]

I really don't understand why you need the limitation that time-series must have a frequency consistent?