Time series forecasting - running in a loop

I have used the KNIME MLP workflow for time series forecasting and modified it to use my own data for forecasting. My input values for forecasting depends on a variable. This variable when used to query a database gives me the set of values that acts as an input for the forecast. However, I have multiple such variables in the database that returns multiple sets of values to be forecasted. If I have to run the knime workflow in a for loop for example and it loops the number of times based on the number of variables in a text file - is there any way to achieve this?


Hi Sachin,

EDIT: @sachinwagh A colleague has pointed out to me, that I might have overcomplicated what you are interested in. The same result can be achieved by extending your workflow and wrapping the processing of a single data slice inside a loop. Then you can loop over all relevant values of the variable inside your master workflow and do not need to cal one workflow from another.

yes, one can achieve this by using Call Local Workflow (Row Based) or Call workflow (Table Based). See an example here or here, respectively.
In the first scenario one would provide the table containing the list of parameter values as input, each row would be captured by the callable workflow and query the DB to extract a data subset and do processing. In the second you would have a loop in the master workflow, query the DB in the loop, get the data in a table and call the second workflow to do modelling on it.


Thanks a ton Misha, will work out both the examples and select the best fit.