Icon

Rolling Time Series Predictions

This workflow demonstrates how a recursive loop can be used to do rolling predictions, i.e. use some existing data to bootstrap the prediction and then use the predicted values themselves as features for the prediction of the next point in time.

Train a simple model Prepare some example data 200 daysrandom targetCollect only predictions,pass back last 10 rowsLag target10 daysLearn modelPredict targetOld dataand new rowOnly dateand targetLast 10 rowsWe don't needthe firstrow anymore+1 day withlagged valuesfrom previous days Create Date&TimeRange Math Formula RecursiveLoop Start Recursive Loop End Lag Column PolynomialRegression Learner RegressionPredictor Concatenate Column Filter Row Filter Row Filter PreparePrediction Row Train a simple model Prepare some example data 200 daysrandom targetCollect only predictions,pass back last 10 rowsLag target10 daysLearn modelPredict targetOld dataand new rowOnly dateand targetLast 10 rowsWe don't needthe firstrow anymore+1 day withlagged valuesfrom previous daysCreate Date&TimeRange Math Formula RecursiveLoop Start Recursive Loop End Lag Column PolynomialRegression Learner RegressionPredictor Concatenate Column Filter Row Filter Row Filter PreparePrediction Row

Nodes

Extensions

Links

Download

To use this workflow in KNIME, download it from the below URL and open it in KNIME:

Download Workflow
Created by: alexander
Created at: 2020-02-03
On NodePit since: 2020-02-03
Last update: 2024-04-24
Created with KNIME version: v4.1.0
Tags: Time SeriesPredictionRecursive Loop

Deploy, schedule, execute, and monitor your KNIME workflows locally, in the cloud or on-premises – with our brand new NodePit Runner.

Try NodePit Runner!