Training a Gradient Boosted Forest
To classify the four different audio signals so that we can tell them apart, we’ll need to do more than just apply the Fourier transform; we need to build a model on our cross-sectional data. Training a Gradient Boosted Forest in KNIME is very easy. We’ll use the Gradient Boosted Trees Learner node, which only has a few configurable options that we’ll concern ourselves with. We’ve chosen to use the Gradient Boosted Forest model due to its ability to handle high-dimensional data and its impressive predictive power.
Applying the Fourier transform in KNIME
The workflow we’ll use to train the model and do all of our preprocessing with the Fourier transform can be seen in Figure 8.11. You’ll notice it is a small workflow with some of the binning logic placed inside the FFT and Binning component.
Figure 8.11 – Training workflow with FFT preprocessing
When constructing this...
