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.

I trained a neural network in MATLAB but R value is lying between 0.2 and 0.3. Any tips on how to make the R value close to 1?

While using ANN tool from MATLAB, the value of R is the correlation coefficient. The correlation coefficient 𝑅 is used together with the 𝑀𝑆𝐸 as performance indexes of the ANN. The 𝑅 values vary between -1 and +1, however, if the 𝑅 value is close to +1, a high network performance and a positive linear relationship can be

achieved. The weights and bias of neurons inter-connection are adjusted during training until the network learns the correlation between input and output variables, i.e. until the minimum mean square error (MSE) is obtained.

The ANN models used two different basic backpropagation (Bp) algorithms that are basic Bp and Bp with momentum and learning rate coefficients. These indicators are root mean square error (RMSE), the mean absolute percentage error (MAPE), mean absolute bias error (MABE), correlation coefficient (r), and coefficient of determination (R2). Accurate ANN models were based on the minimum values to RMSE, MABE, and MAPE, and maximum values of r and R2.

R is the coefficient of correlation and R2 is the coefficient of determination.