Doesn't look like you need anything fancy here. Load you time serie in a dataframe, and: > - Variability: how "spread out" are the data points from one anothe

Doesn't look like you need anything fancy here. Load you time serie in a dataframe, and: > - Variability: how "spread out" are the data points from one another? So basically df.std(), with rolling variants for short term / long term. > - Trend: direction of data path, up or down? - Slope: are the data points increasing or decreasing? Just do a simple rolling linear regression of your data point against time.