Azure Auto ML forecasting - dependent variables

Rajeshkumar Mourya 6

Hi, I am trying to use Azure auto ML for forecasting. The dataset has a date_time column, a target variable and other columns that affect the target variable. I have deployed the model as a web service. But I am finding it hard to use the service/model for forecasting future frames. Let's say I need to forecast for the next 4 hours (data frequency is 5 minutes), but the model is asking for other column inputs as well. Can you please help me to resolve this?
TIA,
Rajesh

1 answer

Ramr-msft 17,616 Reputation points

2021-02-16T16:33:58.797+00:00

@Rajeshkumar Mourya Thanks for the question. Can you please add more details about the use case that you are trying. Here is the docs/samples for Automated ML forecasting enables businesses to forecast revenue, inventory, sales, or customer demand. Customers can run Automated ML experiments by using a no-code UI experience or a code-first Python SDK experience and sample Jupyter Notebooks depending on their expertise.
Documentation: HTTPS://AKA.MS/AUTOMLFORECAST
Please sign in to rate this answer.
Rajeshkumar Mourya 6 Reputation points

2021-02-16T22:44:34.75+00:00

Thank you for responding. I am trying to predict the response time of my system in the next 4 hours. The dataset has a date-time column and data is generated every 5 minutes. other columns are CPU time, Database calls processing time, system id, free memory and processing units available. These columns affect the response time of the system. I have trained and deployed the model using AutoML UI experience and have a scoring URL generated through it. Next step is I try to consume this model and have a test data set with 129 entries, the model is consumed through Azure python SDK. I get the results for the 129 entries but there is no future forecast included.
How do I retrieve forecasted response time values from the deployed model? Am I missing anything here?

Rajeshkumar Mourya 6 Reputation points

2021-02-16T23:05:03.057+00:00

Datatypes details are as following:
CPU time - decimal
Db call time - decimal
free memory - numeric
processing unit - numeric
system-id - string

Ramr-msft 17,616 Reputation points

2021-02-23T04:11:46.99+00:00

@Rajeshkumar Mourya Thanks for the details. Can you please share the sample dataset to check. Please follow the doc for prerequisites. Here is the link to the sample for Rolling Prediction & Evaluation.
Business requirements often lead to analysts and data scientists predicting further into the future on less historic data than recommended. To help users get sound predictions, we have released rolling prediction and evaluation which allows users to forecast out shorter periods, automatically appending these predictions to the training data, and forecasting until the desired forecast is reached.

https://github.com/Azure/MachineLearningNotebooks/blob/master/how-to-use-azureml/automated-machine-learning/forecasting-forecast-function/auto-ml-forecasting-function.ipynb

Rajeshkumar Mourya 6 Reputation points

2021-02-23T04:46:29.443+00:00

@Ramr-msft Thank you for the response. Please find attached a sample of the dataset. Thank you for sharing the link for the forecasting function, I will go through the details.
Cheers,
Rajesh
Sign in to comment