Hi Fir,
The "newff" function has been obsolete for years and has been removed from newer versions of MATLAB. When you type “>> doc newff” in the command window, you will see the following message in the documentation:
“Obsoleted in R2010b NNET 7.0. Last used in R2010a NNET 6.0.4.”
The "newff" function was used to create a feedforward backpropagation network. The current recommended function for feedforward backpropagation is “feedforwardnet”.
You can follow the steps below to perform cross-validation with a feedforward neural network for the Fisher Iris data:
- Data Preparation: Convert the iris measurements into input features and encode the species labels numerically (using one-hot encoding for neural networks).
- Partitioning: Split the dataset into "k" folds (e.g., 5-fold cross-validation).
- Training and Testing Loop: For each fold, train the feedforward neural network on the training portion of the data and use it to predict the species on the test portion. After making predictions, compare them to the true labels and calculate the misclassification rate for that fold.
- Averaging: After all folds, average the errors to estimate the network’s cross-validation error.
You can read more about "feedforwardnet" by going through the following documentation link:
I hope this helps.
