I collected data from a motor at different RPMs creating different labels for a balanced state.
Similarly, collected data at same as previous RPMs by creating imbalance in the motor.
At each RPM, I have collected 10s data at 100 Hz sampling frequency 5 sets in balanced condition and unbalanced condition.
I trained the network using spectral features, but it took 18 classes instead of 2 ( each having 9 RPM states with 5 sets of data ) leading poor accuracy.
Now i would like to merge the data of all RPMs belongs to
You can rename your labels as “good” or “bad” either using the UI as shown below, or by writing a small script using the API (list all samples and edit labels: https://docs.edgeimpulse.com/reference#editlabel)
However if you don’t have that many samples and you have them stored on your PC, the fastest way can be to remove all your existing samples (Dashboard -> Delete all data), and re-upload them using the labels "good"or “bad”.
Re-upload. You can also go to Dashboard > Export to download the original files, then go to Data Acquisition > Upload to upload them again and then set a label.
