topic Re: Time Series Forecasting for Multiple Customers in Qlik AutoML / ML Experiment in Qlik Predict

Time Series Forecasting for Multiple Customers in Qlik AutoML / ML Experiment

sarak2025 — Wed, 06 Aug 2025 18:51:56 GMT

Hi,

I’m working on building a time series forecasting model to predict Gross Profit for the next 18 months (Aug 2025 through Dec 2026), and I’m looking for expert guidance on the best approach within Qlik’s ML environment—especially for handling multiple time series (one per customer).

Dataset Details:

Time Range: Jan 2022 – July 2025 (43 months of historical data)
Columns: CustomerID, EOM (End of Month), Amount (Gross Profit)
Scale: 500+ unique CustomerIDs
Goal: Generate individual forecasts per customer for the next 18 months

Below is a screenshot of the dataset.

I’m specifically exploring the "ML Experiment" feature and wondering:

Is there a tutorial or example that shows how to structure a time series forecast for multiple entities (like CustomerID)? I’ve reviewed the documentation but haven’t found a clear example of this use case.

If you’ve implemented something similar or know of any Qlik-native resources, tutorials, or best practices, I’d love your insights.

Thanks in advance!

Re: Time Series Forecasting for Multiple Customers in Qlik AutoML / ML Experiment

marksouzacosta — Thu, 14 Aug 2025 12:19:37 GMT

Hi @sarak2025,

Maybe @igoralcantara can help you on this.

Regards,

Mark Costa

Re: Time Series Forecasting for Multiple Customers in Qlik AutoML / ML Experiment

igoralcantara — Thu, 14 Aug 2025 12:49:56 GMT

@sarak2025

Qlik Predict at the moment does not support Time Series. This feature is coming soon, maybe late September or early October. If you want to know what's coming, I wrote an article about it.

You can read it here: Beyond the Event Horizon: Multivariate Time Series with Qlik Predict

What you can do in the meantime is to create a regression model with some feature engineering. On top of those features you have, you can add Rolling7DaysAverage, Rolling14DaysAverage, Rolling30DaysAverage, and other features that might make sense.

However, if what you want is just to forecast a trend line, I wrote a couple articles that discuss just that:

How to build Trend Lines and Forecasting in Qlik

Fourier Transform in Qlik: Separating Signal from Noise

Re: Time Series Forecasting for Multiple Customers in Qlik AutoML / ML Experiment

sarak2025 — Thu, 14 Aug 2025 16:17:46 GMT

Hi all, thank you for your reply!

I spoke to the sales team from Qlik earlier this week and they mentioned

"Qlik currently supports multivariate time series forecasting through Qlik Predict, formerly known as Qlik AutoML. This capability is fully available in Qlik Cloud and allows users to forecast trends across multiple variables, enabling more accurate and context-driven insights using automated machine learning."

I will have a demo with them tomorrow. Hopefully this "new" feature is just added without us knowing. I will update as I have more information 🙂

Re: Time Series Forecasting for Multiple Customers in Qlik AutoML / ML Experiment

igoralcantara — Thu, 14 Aug 2025 16:28:00 GMT

I am pretty sure it is not yet available. It is still being tested. Maybe they are referring to a line chart forecast line, which is not Time Series.