more statistics gone astray.

i spent the majority of my weekend grading stat papers on regression. again. it was exactly as glorious as it sounds. here are some of the highlights. first, i'm not sure why a number of people were so fatalistic:

• After running almost a “billion” regressions...
• I got lost in the models to be honest.
• The histogram is miserable.
• This was a model out of desperation.
• I can’t simply unlog the error of the forecast.
• I do not like this regression (but I'm running out of time and it’s the best model that I can make, sadly so, that has ANY basis in reality).

Some were mildly more optimistic but still very, very confused:

• The only issue I see with this model is the poor histogram of the residues.
• It also made sense that the price was mainlolky depend on the year.
  
• But it’s only slightly worst.

There were also a few statements made courtesy of captain obvious and his sidekick, bullshit dog:

• [I'm using the] number of years since 1971, starting in 1972.
• At first glance, the final model may not make sense. However, upon further review, the final model seems reasonable. [taken verbatim, btw]
  
• That’s not exactly useful for making business decisions. Throwing a rock at the broad side of a barn might be more precise.
  

Finally, i'm not sure what else these folks had on their minds, but i'll preface with the clarification that one of names used in the case was HEM FIR, not:

• Hem Firm
• Hemi Fur
• Hemp Fir