I performed a standard error on my machine learning model to predict protein structure. The graph Im showing here is a snippet of the actual data and I deleted some irrelevant info. The y axis is the fractional total abundance and the x axis is the the actual GCMS, the model prediction in replicates.

My question is if the standard error lines (gray lines in the model prediction). How do I explain to my PI that since they overlap with the actual GCMS data, it should be fine. He is arguing that its abundance mis-predicted by a factor of at least two (and the errors go in both directions). My image is not to scale the original one is.

Isn't it that the overlap of SE with the actual data just means its not statistically significant?enter image description here

  • $\begingroup$ what is pred. 1 and pred 2 ? $\endgroup$ Mar 3 at 10:41
  • $\begingroup$ How did you compute standard error ? Also, what is GCMS ? $\endgroup$ Mar 17 at 13:53


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.