Sales Forecast with recursive loop

[Patrick92]

Hello Dear Forum,

i need your help and I hope you can understand my question.

I have sales data from 2013 - June 2015 and should make a forecast in sales for July 2015.

Now I thought about to do this best with a recursive loop but unfortunately I’m lost to this one.

I hope somebody can help me to create a easy and understandable workflow for my problem.

Many thanks in advance!

[Snowy]

Creating a forecast… Particularly, choosing the right method to forecast… Depends quite a lot on the goals of the project, available data, how accurate it needs to be, whether external factors might influence the forecast, etc.

If you can provide a little more context to what you have at your disposal, someone can probably help you.

[mlauber71]

@Patrick92 welcome to the KNIME forum

Like @Snowy said: it would be helpful if you could give us some more details.

Then. We had a somewhat similar discussion about the prediction of numeric targets (unemployment in the UK) here:

Can KNIME be used to show the employment and unemployment rate in the UK from the year 2008-2018, showing the prediction for the next year?

I set up an environment that first transforms you data from 2008-2016 into a table that has the Area (region) as an ID since I understand that is the unit you would want to predict. [image] The Target is taken from 2017 “Emp UK percent”. If you want Unemployment you would have to change that. The years are transformed into columns, while the first year of the data becomes _0 and the year before that _1 and so on. [image] Then I construct a Test dataset using the data of past 9 years from …

This resulted in a collection of workflows (please consider the whole workflow group):

compare several regression models on historical unemployment data – mlauber71

Or this:

Unsupervised machine learning (gradient boosting regression)

I set up an example that should look like your data (if not feel free to adapt and inform). We have two years of data (from one city) and data from January and want to predict the coefficient for Feb 2020. [image] 2019 serves as training and 2018 as test data. You might have more complex settings with cross-validation if you want. And also since you have the data you might provide coefficients of some more past months. Each line represents its own time series so to speak. Of course, such a se…

Then there is this collection about machine learning methods. In your case a regression model might be the right one. Or you might have to use something to predict based on time series (no collection for that, yet):

Machine Learning Meta Collection (with KNIME) – mlauber71

Often it does make sense to familiarise yourself with several methods and especially think about your business case and the data you have available and if what you want to find is somewhere in the data you have.

[Patrick92]

Thank you for your quick responses.

Basically its this competition, who I should solve with a linear regression and a loop.

Rossmann Store Sales

The task is to predict the sales from each store for Juli 2015 (next 31 days).

We should only concentrate at the “train” and “test” data. I cant upload the both data because they are .csv-Data but you can have an access at the link I posted.

I tried to create a Workflow but somewhere there is an error. The Workflow is attached and a screenshot also.

I think a recursive loop is not necessary. The Workflow should be easy and understandable but also exactly as possible.

Thank you all for your help and time!

Rossmann_Sales.knwf (31.7 KB)

[Daniel_Weikert]

Can’t execute your flow as I don’t have the data but if you filter out the dates
$Date$<=“2015-06-30”
in your first row splitter then I am not sure why you do the same for the second spiltter
$Date$<=“2015-06-30” => TRUE

[Patrick92]

Thank you for your help!

You can download the data from the kaggle link I posted if you want to execute the workflow.

You may right the one date node is redundant but the workflow still doesnt work.

You see I’m a total beginner thats why I need help. Nevertheless I’m ready to learn.

[Patrick92]

Nobody who can help me? 🥲

[Daniel_Weikert]

What’s the error message. I can only see a warning in your flow
bR