Hello. I signed up to these forums trying to find a solution to this... relatively simple problem.

I'm trying to calculate the Root-Mean-Square Error of predicted and observed models (wind roses to be specific). The models each have several data points, and each data point has several values and frequencies.

Code:

data_point_1:
observed_frequency: 100 050 010
observed_value: 0.5 0.3 0.5
predicted_frequency: 090 060 005
predicted_value: 0.4 0.5 0.3
data_point_2:
...

Disregarding the frequencies this is an easy thing to solve.

Code:

RMSE = (Sum((observed_value_p_i - predicted_value_p_i) ^ 2) / n_count_values) ^ (1 / 2)

But the data points all have different frequencies on both the observed and predicted models (wind doesn't blow evenly in all directions), I need to weigh the Error based on the observed and predicted frequencies, and this part has me a bit stumped. Do any of you have an idea how to accomplish this?