I am working on a classification problem and i have a dataset consisting of two classes (error, no error) with many other parameter columns. I am trying to build a remaining useful life model but lack information in my dataset for transitional points(last good point before entering error phase, first bad point entering error phase). i would like to create synthetic data that mimic these states i.e have more no error values but that increasingly turn towards the error values and have error values that slowly increase towards the extreme errors state. So i need transition values, not only extreme values. My column parameter values are of various types, time series, constant, digital and so on.
All that i see online like sdv and timeseries-generator seem to only create synthetic data of the same range as real data, which won't provide me with this transitional data from no error to error which i really need. I would really appreciate it if someone could guide me through steps or perhaps refer me python packages/methods that would help me create such a dataset. Thank you.
for now i can have two different datasets with no error parameter values only and error parameter values only. But both datasets were recorded at this extreme/exact conditions. I need transitional data. Thank you