The values in the training dataset are integers and distributed in the range of [-15871; 16213] (please note, that the ranges are very different for the different sequences, but we estimate the largest range for all sequences). Thus, they may be encoded as the 15-bit variables (1 sign bit, and 14 data bits representing a number upto 16384)
sc_train.exe D:\Data\Raw_6_sensors 6 16 1 512 2000 1000 0.01f D:\dictionary.dic
The resulting features are represented by floating point values in range [-1; 1].
sc_test.exe D:\dictionary.dic D:\Data\Raw_6_sensors_examples\S1-ADL1-1\ex2_6-69.txt 6 16 1 D:\ex2_6-69.dat