When the R value is computed manually using formula or using tools such as MS Excel or SPSS, the entire dataset is used to build the empirical relation between the inputs and output.

So, logically All R value should be used, as it represents the entire dataset.

With Regards

Ashu

coefficient of determination.

hi Tarke

Coefficient of correlation is R and coefficient of determination is R^2 .Usually often, to compare the proposed model with traditional equations, validation stage of data should be to select.

R value, the coefficient of correlation or determination are used to measure the correlation between actual and predicted value. it measures the direction and strength of the linear relationship between actual and predicted value.

You may considered all R for comparison of training , testing and validation data.

If you want to consider any one then u go for validation error.

When the R value is computed manually using formula or using tools such as MS Excel or SPSS, the entire dataset is used to build the empirical relation between the inputs and output.

So, logically All R value should be used, as it represents the entire dataset.

With Regards

Ashu

In order to find network performance, regression All is preferred, i.e., testing should be done on the large data. Normally, to decide the network performance, MSE, MAE and R^2 are measured as performance metrics of ANN models and not just regression R.

For comparison: The same data set should be used to calculate R-value manually using the formula for the coefficient of regression.

Hi Prajot Tarke

(r) is a Coefficient of correlation and (R2) is a coefficient of determination.

Coefficient of correlation is R and coefficient of determination is R^2

Dear Prajot,

R is a coefficient of correlation and R² coefficient of determination

Good luck

All values of R (for testing,training and validation) are used.The closer the R value to 1 the better the correlation.

R is a coefficient of correlation and R^2 is a coefficient of determination

The most the neural network measure is considered both the R and the MSE or MAE.