Currently, I developed using multiple regression, I developed statistical prediction and I obtained the the result, then I want to go to verification using tercile probability. So, I like to calculate the the probability
a. Below Normal (BN): using the value of less than 33% as threshold
b. Normal(n) : using the value of range 34%-67% as threshold
c. Above normal (AN): using the value of greater than 67% as threshold
Here is the random data
obs=c(28,72,45,52,87,25,89,104,65,96,74) which is an observation
prd=c(32,56,58,98,29,75,98,81,92,85,78) which is predicted
I like to obtain P(AN), P(N), P(BN) for each year /data
I hope to get your help soon.
Dear Andrey, I define the threshold as follow for each of them
a. P(BN)-- Obs
Well, now it is important that you formulate your task precisely.
1) What type of data use use as input (do you use only obs and prd)
2) Give an example of what you want to see as a result.
I think you probably would need to regress prd on obs and then obtain your probabilities based on the posterior conditional distribution for obs.
In other words, you will need to estimate the following regression model: obs=f(prd)
When you have the conditional PDF for obs given prd, you can find your probabilities.
- Badges
- Science topic