Hyperparameter tuning CNN

Hyperparameter tuning CNN

hp_tuning_CNN(
  data,
  train_size = 0.8,
  random_seed = 1234,
  n_steps,
  n_filter,
  neuron,
  learn_rate,
  epoc
)

Arguments

data

Dataset to split and tuning

train_size

Proportion of the x to split as train subset. By default the value is 0.8

random_seed

An integer to control the random number generator used. By default the value is 1234

n_steps

Number of steps to split into samples

n_filter

Number of filters in the convolutional layer

neuron

Number of neuron in the dense layer

learn_rate

Learning rate in the compilation with Adam optimizer

epoc

Number of epochs in the training model

Value

Returns: Dataframe with the combinations of the parameters and the calculated error Better combination of hyperparameters that minimizes the train and test error

Author

Catherine Rincon, catherine.rincon@udea.edu.co