Hi,
I have measuring data from a test rig. 2 columns. first column force and the second column distance .
I would like to have at 5 distance values (10mm 20mm etc.) the Y-values (forces). The data has a big sampling rate so I have to work with borders (for example 10mm → 9.8 to 10.2).
I try to do it with 5x row filter + math formula (mean) + concatenate.
Has everyone an easier way to do it with KNIME.
Thanks in advance!
Hello @HofmannO
You are describing a challenge that could be tackle with different methods. Even you may have to explore your data aiming to know linearity of the data.
From my perspective this is a ML ‘Regression’ challenge. Where you calibrate the model with your test data; then feed an additional Predictor with your discrete selected sampling, once you’ve checked the scorer of the model. From you testing data set description, you could achieve better scores by using Polynomial Regression or XBoost (Regression) nodes.
Other approach would be by coding -with classical computing for two variables-, you can plot your adjusted Linear, Polinomial, Exponential, Logarithmic… regression plot, and get back the the adjusted equation (testing the best adjusting method r2); then apply the regression function to your discrete reporting values.
Any other method would be interpolative, that can be inaccurate. But it depends on your test data and reporting requirements.
Hello,
Thanks for your answer. very interesting solving the problem with a regression. But for me it feels a little bit more complicated. the data has a little scattering so I have to set up a moving average (in excel 40 points to 1 average point) for it. In the first try I filter the desired points. For each point I must use a row filter and afterward I have calculate a mean value. So actually it feels not comfortable if I have to chance points or add some points.
Maybe your way is still more easier, having a equation set the x component to get the Y result. I will try.
Thanks in advance!