Wednesday, July 27, 2011

Building A Model For Attrition

Round 3. Let's determine what factors we want to use to build a better model for attrition. This third approach will use actual climatic values for a given day to calculate attrition rather than arbitrary designations for season or region. Values will be randomly determined using bounds from the Köppen-Geiger climate classification and/or world climate data.

Attrition will be broken into two components, climate and route conditions. Within each component, smaller cumulative elements will sum to a total factor, then each factor will be multiplied by a baseline value (yesterday's attrition value). I'll get into the math in a following post.

Climatic Factors

Humidity - Low humidity (especially combined with cold air) will dry out the throat and lungs, causing respiratory stress and increasing susceptibility to disease. High humidity reduces the body's ability to cool, compounding heat stress.

Temperature - As a body heats up, it suffers from a multitude of debilitating and progressively fatal symptoms related to heat stress. Lower temperatures are dangerous to a lesser degree than higher temperatures, but will eventually lead to death all the same.

Storms and Frontal Changes as represented by Precipitation - This will be a catch all for storm activity in general, from a hailstorm to snow accumulation and sudden changes in weather conditions.

Major climatic factors sourced from Climate effects on human health, Kalkstein, L. S., and K. M. Valimont, EPA Science and Advisory Committee Monograph no. 25389, 122-52. Washington, D.C.: U.S. Environmental Protection Agency.

Pardon the horrible citation, I can get away without using MLA or APA style here :D


  1. Good luck. I'd love to see you succeed.

  2. Thanks. I'll post the code too so others can fiddle with and use it if they would like to.

  3. Does the "frontal changes" include blowing wind? Strong winds can affect the character pretty much.

    call Nepal

  4. I'm having a hard time trying to capture elements for attrition beyond temperature and humidity. There are plenty of effects that can certainly cause attrition, like driving storms, desert sandstorms, hail, sudden changes in temperature, deterioration of roads from weather events, tornadoes, lightening strikes, etc. But I am having trouble capturing them in a systematic and quantifiable manner.

    Temperature is pretty easy, the human body pretty much shuts down at 35 to 40 degrees due to thermodynamic law. The lower end of temperatures are not so definite, but easy to grasp.

    What about a storm? Should I generate wind speeds, precipitation, rate of rainfall, and amount of ground cover to assign an attrition value to a storm? Even 0.1" of rainfall can cause significant damage in the form of hail, but a light drizzle of 0.1" would be trivial.

    So I guess that is a long way to answer your question Jimmy. I am actually considering removing storms and precipitation from the climatic portion of attrition and placing it elsewhere. But that is a post that I have yet to find the time to make...