Inference on training data impossible??

Hi,

I want to detect potential model overfitting for Azure AutoML forecasting. Usually, I would tune a model on train and then evaluate separately on train and out-of-time test set. But it seems like I cannot do backtesting with forecasting models on train?? That doesn't seem right.

Am I missing something here?

未指定任何檔案、筆記本、測試或進入點。首先重現 issue 中描述的 Azure AutoML forecasting 工作流程，並確定是否支援對訓練資料進行 inference 或 backtesting；完成的標準是記錄該行為，或確定一項具體的 notebook 或 SDK 變更。

由索引模型根據 Issue 內容生成。