Loss and Loss Functions

I know my dear wife has been wanting to blog about her grandfather's funeral that we attended in MI at the end of February. The whole trip has been on my mind since we came back home, and since she hasn't yet had time to blog, I'll offer the following meager reflections and experiences from that trip, as perhaps only a statistician could.

--Mr. Pickles

-------------------------------------

... using the following loss function ...
... we achieve the minimum under the loss function defined by ...
... in these conditions the most appropriate loss function ...

In statistics a loss function is an equation that helps you measure how much you know about a given situation. You make an estimate, and then you measure how far away that estimate is from the true value. The loss function allows you to measure how much information you loose with a particular estimate. The really important aspect is that a loss function measures something definite.

It's a very important concept in my discipline as a statistician. We are by training a mostly empirical group. Our emphasis is on what can be demonstrated logically, what can be shown with data, and what is quantifiable. As the American Statitical Association says on one of their t-shirts, "In God we trust, all others bring data."

At the end of February I took a break from that discipline and traveled to Michigan with my family for the funeral of Mrs. Pickles grandfather. It was a beautiful funeral. He was a veteran of WWII and a member of the Knights of Columbus. The Knights were dressed in their regalia -- swords at the ready to salute their brother. There were also two soldiers who draped his coffin with the American flag, and they ceremoniously folded it as the familiar melody of "Taps" rang through the air. It was very moving, but for me the most moving part occurred before the funeral.

We arrived at the funeral about 20 minutes early for the viewing, and I saw my wife's grandfather lying there -- in words perfectly expressed by Miriam -- looking as though "he were asleep." I half-expected him to sit up at any moment and say something in what I remember to be his somewhat creaky yet jovial tone. I don't have many memories of him. Most of what I have are second-hand from my wife and in-laws -- about how much they loved him, how much he deeply loved his wife, and about what a good man he must have been to raise a wonderful and daring man (and nut) like my father-in-law.

I stood there looking at him with my expectation for what would be a resurrection. In knowing that it would not happen I experienced a much different sense of loss from the one I know through my work. It was something else--something undefinable and unmeasurable. I can barely describe it now as I type this. All of the positivist impulses of my profession and daily practice were rendered meaningless. I came face to face with something far more real than any of data I work with on a daily basis, but I couldn't touch it, see it, or quantify it. It was a true sense of loss, and a knowledge that our world was a better place with him, and it is a lesser place without him.

Please pray for his soul, so that he will, in turn, soon pray for ours. Requiem in pace.