The Time Series prediction Problem

Time series prediction requires the prediction of a value at time t, x(t), given its past values, x(t-1), x(t-2), …, x(t-n). How do you implement a model for time series prediction in KNIME? For time series prediction, all you need is a Lag Column node!

For example, I have a time series of daily data x(t) and I want to use the past 3 days x(t-1), x(t-2), x(t-3) to predict the current value x(t). This is an auto-prediction problem. Introducing exogenous variables, like y(t) and z(t), into the prediction model, turns an auto-prediction problem into a multivariate prediction problem. Let’s stick with auto-prediction. What we will build is easily extendible to a multivariate prediction.

Over the past years, KNIME has grown from being an interesting new tool for data analysis to a platform for analytics that embraces openness at its core principle. In contrast to proprietary software vendors, KNIME allows integration from any source or vendor, transparency creating reusable workflows, collaboration both inside and outside of your organization, and agility for quick data exploration. The result is faster, more powerful insight into complex data. You can read more about the five pillars upon which an open analytics platform stands on here. This openness is something we, and many of our users, strongly believe is the future of data analytics.