master of none in datascience
Attention: Still in experimental state
pip install zeugAssume you have cyclical data like weekday. 0 for monday and 6 for sunday. Even the difference between sunday and monday is only 1 day the numerical difference between 0 and 6 is big.
weekdays = [0, 1, 2, 3, 4, 6]
weekdays_sin, weekdays_cos = zeug.sin_cos_transformation(weekdays, period=7)We can use sinus and cosinus tranformation to preprocess cyclical data like hours, weekdays and months. The result consists of the sinus and cosinus values that hold the information about the cyclical behavior.
