Regression output interpretation

[Sofia Pinto]

Hi,

I've been using randomJungle for both classification and regression but I'm having some trouble interpreting the output files for the regression problem.

The independent variables are SNPs and the dependent variable is a numeric measure of a phenotypic trait. Bellow is an example of a .confusion file obtained:

Number of variables: 275897
Test/OOB set:
Variance of y^ (SS_Residual) = 0.0257776 (18.0701)
Variance of y (SS_Total) = 0.0249509 (17.4906)
Explained Variance (SS_Predictor) = -0.000826715 (-0.579527)
Accuracy estimate = R^2 = SS_Predictor / SS_Total = -0.0331336
Error estimate = 1 - Accuracy estimate = 1.03313

My questions are:

1. What are the two values in the three variances results, for example 0.0257776 (18.0701)?
2. What is the meaning of negative values for the explained variance and the coefficient of determination (R^2)?
3. If I would like to know how much of the variance of that trait is explained by the SNPs which value should I take?

Thank you for your attention.

Best regards,
Sofia