{

"session_name": "test_script 2",

"session_description": "DL testing for UI integration",
"design_state_data": {
"target": {
"prediction_type": "multiclass",
"target": "species",
"target_treatment": "Error"
},
"train": {
"time_ordering": false,
"time_variable": "",
"time_ordering_type": "ascending",
"sampling_method": "No sampling",
"number_of_records": 0,
"random_percent_to_use": 60,
"rebalance_method": "SMOTETomek",
"train_ratio": 0.8,
"random_seed": 47
},
"metrics": {
"optimize_model_hyperparameters_for": "AUC",
"optimize_threshold_for": "F1 Score",
"compute_lift_at": 0
},
"feature_handling": {
"sepal_length": {
"feature_name": "sepal_length",
"is_selected": true,
"feature_variable_type": "numerical",
"feature_details": {
"numerical_handling": "Keep as regular numerical feature",
"rescaling": "No rescaling",
"missing_values": "Replace with",
"impute_with": "Median of values",
"impute_value": 0,
"threshold_type": "Average of values",
"threshold_value": 0
}
},
"sepal_width": {
"feature_name": "sepal_width",
"is_selected": true,
"feature_variable_type": "numerical",
"feature_details": {
"numerical_handling": "Keep as regular numerical feature",
"rescaling": "No rescaling",
"missing_values": "Replace with",
"impute_with": "Median of values",
"impute_value": 0,
"threshold_type": "Average of values",
"threshold_value": 0
}
},
"petal_length": {
"feature_name": "petal_length",
"is_selected": true,
"feature_variable_type": "numerical",
"feature_details": {
 "numerical_handling": "Keep as regular numerical feature",
"rescaling": "No rescaling",
"missing_values": "Replace with",
"impute_with": "Median of values",
"impute_value": 0,
"threshold_type": "Average of values",
"threshold_value": 0
}
},
"petal_width": {
"feature_name": "petal_width",
"is_selected": true,
"feature_variable_type": "numerical",
"feature_details": {
"numerical_handling": "Keep as regular numerical feature",
"rescaling": "No rescaling",
"missing_values": "Replace with",
"impute_with": "Median of values",
"impute_value": 0,
"threshold_type": "Average of values",
"threshold_value": 0
}
},
"species": {
"feature_name": "species",
"is_selected": true,
"feature_variable_type": "categorical",
"feature_details": {
"categorical_handling": "Frequency encoding",
"missing_values": "Drop rows",
"impute_with": "Most frequent value",
"impute_value": "",
"drop_dummy": "Default",
"min_frequency": 1,
"max_categories": 50,
"search_string": ""
}
}
},
"feature_generation": {
"linear_interactions": {
"linear_interactions": [],
"impute_with": "Average of values",
"impute_value": 0,
"linear_scale": "robust"
},
"polynomial_interactions": {
"polynomial_interactions": [],
"impute_with": "Average of values",
"impute_value": 0,
"polynomial_scale": "robust"
},
"explicit_pairwise_interactions": {
"explicit_pairwise_interactions": [],
"explicit_pairwise_scale": "robust"
}
},
"feature_reduction": {
"feature_reduction_method": "No Feature reduction",
 "No Feature reduction": {
"is_selected": true,
"num_of_features_to_keep": 5
},
"Correlation with target": {
"is_selected": false,
"num_of_features_to_keep": 3
},
"Tree-based": {
"is_selected": false,
"num_of_features_to_keep": 3,
"depth_of_trees": 2,
"num_of_trees": 10
},
"Principal Component Analysis": {
"is_selected": false,
"num_of_features_to_keep": 3
},
"LASSO regression": {
"is_selected": false,
"num_of_features_to_keep": 3,
"regularization": []
}
},
"hyperparameters": {
"search_method": "Grid Search",
"No Grid Search": {
"is_selected": true,
"random_state": 47
},
"Grid Search": {
"is_selected": true,
"random_state": 0,
"max_iterations": 0
},
"Randomized Search": {
"is_selected": false,
"random_state": 0,
"max_iterations": 0
},
"Bayesian Search": {
"is_selected": false,
"random_state": 0,
"max_iterations": 0
},
"cross_validation_strategy": {
"strategy": "StratifiedKFold",
"StratifiedKFold": {
"is_selected": true,
"random_state": 1000,
"n_splits": 3,
"shuffle": false
},
"KFold": {
"is_selected": false,
"random_state": 1000,
"n_splits": 3,
"shuffle": false
},
 "StratifiedShuffleSplit": {
"is_selected": false,
"random_state": 1000,
"n_splits": 3,
"test_size": 0.1
},
"ShuffleSplit": {
"is_selected": false,
"random_state": 1000,
"n_splits": 3,
"test_size": 0.1
},
"TimeSeriesSplit": {
"is_selected": false,
"n_splits": 3
}
}
},
"weighting_strategy": {
"weighting_strategy_method": "No weighting",
"No weighting": {
"is_selected": true
},
"Sample weights": {
"is_selected": false,
"weighting_strategy_variable": ""
},
"Balanced": {
"is_selected": false
}
},
"probability_calibration": {
"None": {
"is_selected": true
},
"Sigmoid - Platt Scaling": {
"is_selected": false
},
"Isotonic Regression": {
"is_selected": false
}
},
"algorithms": {
"NeuralNetwork": {
"model_name": "NeuralNetwork",
"is_selected": true,
"code_block": "# Following Modules have been configured and
imported in the backend\nimport torch.nn as nn\nimport torch.nn.functional as F\n\
n\n#
###################################################################################
###############\n# Define the neural network architecture within the '__init__'
method. Method signatures and returns\n# should not be altered. Utilize the
'forward' method for implementing activation functions in each layer.\n#
###################################################################################
###############\n\nclass NeuralNetwork(nn.Module):\n\n # Do not change the
method signature. Input and output features are essential parameters.\n # Modify
the body of this method as needed for defining the neural network architecture.\n
def __init__(self,\n number_of_input_features,\n
num_output_features):\n super(NeuralNetwork, self).__init__()\n\n #
number_of_input_features: The total number of features in the dataframe after
preprocessing.\n # The backend ensures that this value is accurately set
after the preprocessing steps.\n self.number_of_input_features =
number_of_input_features\n\n # num_output_features: The expected number of
output features. For regression, set to 1; for classification,\n # set to
the number of target categories.\n self.num_output_features =
num_output_features\n\n #
##################################################################################\
n # USER MODIFIABLE SECTION: Modify the code below as needed for building
out layers\n #
##################################################################################\
n\n\n # Dropout: % of random weights in each layer that are set to 0. Set
the probability ('p') for\n # zeroing random weights. Example: p = 0.1\n
self.dropout = nn.Dropout(p=0.1)\n\n # out_features : This variable can be
changed by the user\n self.input_layer =
nn.Linear(in_features=self.number_of_input_features,\n
out_features=10) # Value set to 10 as an illustrative example\n\n # NOTE:
Number of neurons in each of the layers can be changed as required.\n #
Ensure that the number of \"out_features\" neurons in the previous layer is the
same as\n # \"in_features\" neurons in the next layer\n
self.hidden_layer_1 = nn.Linear(in_features=10, # Output of the previous layer\n
out_features=50) # Value set to 50 as an illustrative example\n
#self.hidden_layer_2 = nn.Linear(in_features=50,\n
#out_features=30)\n self.output_layer = nn.Linear(in_features=50,
out_features=self.num_output_features)\n\n\n\n\n #
###################################################################################
#######\n # Define the activation functions for each layer. Do not change the
method signature.\n # Users can modify this method as needed for customization.\
n #
###################################################################################
#######\n def forward(self, x):\n x = self.input_layer(x)\n\n #
##################################################################################\
n # USER MODIFIABLE SECTION: Modify the code below as needed for defining
activation functions\n #
##################################################################################\
n\n # Apply ReLU activation to x\n x = F.relu(input=x)\n\n x =
self.hidden_layer_1(x)\n #x = F.relu(input=x)\n #x =
self.hidden_layer_2(x)\n x = F.relu(input=x) # User can change as needed\n\
n # Output layer does not require activation function\n x =
self.output_layer(x)\n\n # For binary/multiclass experiments: Uncomment the
line below to apply softmax activation to x\n x = F.softmax(x)\n\n
return x\n",
"param_grid": "{\n 'NeuralNetwork__max_epochs': [200, 100],\n
'NeuralNetwork__lr': [0.01, 0.001],\n 'NeuralNetwork__batch_size': [16, 32]\n}"
}
},
"session_info": {
"project_name": "Deep learning testing",
"project_id": "127",
"experiment_id": "542274047268950852",
"experiment_name": "deep learning testing 2",
"dataset": "iris_DL_test",
"session_name": "test_script 2",
"session_id": "7",
"session_description": "DL testing for UI integration",
"session_mode": "deeplearning"
}
}
}