Closed hpompom closed 2 years ago
Hi @hpompom The graph you show does not provide statistical evidence for the performance of these estimators - it’s merely a reflection of the numpy seed being used to generate the data behind that particular plot. The seed used in the demo file is the first one, 0, so is not set to make the mixed estimator look better or worse than others. If you have not changed this seed, then I guess the seed behaviour has changed with numpy version. In any case, if you would like to provide some statistical evidence of your claim (by e.g. running the type of analysis in our paper), then I will look into it. Best, Ryan
I run the code of turbocharged demo, but I find I can not to reproduce the figure, and the mixed estimators have a very bad performance. Is there something wrong with the part I'm running?