Version 0.01 is purely based on input-sensor-data (DRME_DRU), without any a-priori assumptions, based on Leys (2013). In short, a very robust outlier detection approach based on noise-normality assumption (with possibility of extension to other distributions).
Here are the results. Red dots are outliers. Pdf can be downloaded.
I think that for pressure and temperature, we could get better results if we take a-priori information into account. I will implement this in v0.02.
Version 0.01 is purely based on input-sensor-data (DRME_DRU), without any a-priori assumptions, based on Leys (2013). In short, a very robust outlier detection approach based on noise-normality assumption (with possibility of extension to other distributions).
Here are the results. Red dots are outliers. Pdf can be downloaded.
I think that for pressure and temperature, we could get better results if we take a-priori information into account. I will implement this in v0.02.
cc @fredericpiesschaert