Boilermaker Forecast Challenge (group project)

Instructions:
Input your numeric forecast values for high and low temperatures, max wind speed, and precipitation amount daily valid for the next day in Lafayette, IN. Forecasts are valid for a midnight to midnight time frame.

You must be signed into your himpactwxlab account to submit a forecast. Forecasts are due by midnight without exception. If you submit a forecast after a midnight, it will be valid for the following day and not the current day. If you make a mistake when submitting, please do not try and resubmit your forecast. Instead, please e-mail the game administrator at fcstgame@himpactwxlab.com with your edits, and your changes will be made for you.

Good luck!


Purdue Forecast Game


Boilermaker Forecast Challenge Rules and Scoring

Forecasts will be collected before midnight valid for the time from midnight to the following midnght. High and Low temperature, maximum sustained wind speed, and precipitation quantity are the values to be forecast. Temperature and wind are compared to observations using absolute error. (Running statistics will also be kept using root-mean-squared error RMSE.) Precipitation forecasts are sorted into bins. The forecast bin is then compared to the bin that the observed value falls into.

Precipitation (in.) bins are as follows:
 bin #
low end 
 high end
 10.00
 Trace
 2 0.010.05
 3 0.060.10
 4 0.110.24
 5 0.250.49
 6 0.50 0.99
 7 1.00 infinity

For example, with forecasts of 0.30 and 0.40 and an observation of 0.45. Both forecasters would receive a perfect score for precipitation because all the values fall into the same bin. Precipitation score is based on number of bins that the forecast value is away from the observed value.

Running scores of RMSE will be kept to create an all encompassing cumulative score:

(0.332871 * RMSE of High Temp) + (0.343081 * RMSE of Low Temp) + (0.214752 * RMSE of Wind Speed) + (0.939336 * RMSE of Precip Bins)


Forecasters will also be ranked against each other for each of the variables. The sum of these ranks will be taken where the forecaster with the lowest sum of ranks would be considered the best forecaster.

For example, someone who was first in max temp, and wind ,and third in min temp and precipitation would receive a sum of ranks of (1+3+1+3)/4 = 2. The same sum of ranks would be assigned to a forecaster that was second for all categories (2+2+2+2)/4 = 2.


Results


Comments