CV10-DNN-TF.py

import random
random.seed(a=1337)
k_fold = KFold(n_splits=10)
LABEL_COLUMN = "Ask.Low.Pc_H4_8AM"

resultz=[]
layerz=[]
stepz=[]

for  i in range(100):
    
    print("EXPERIMENT " + str(i))
    
    resultzz=[]
    
    random.seed(a=1337+i)
    num_layers = random.randint(2,5)
    layer_sizes =[]

    for j in range(num_layers):
        layer_sizes.append(random.randint(2,1024))

    layerz.append(layer_sizes)

    step = random.randint(10,1000)
    stepz.append(step)
    
    j = 0
    
    for train_indices, test_indices in k_fold.split(sellers):
        
        model_dir = './models/m' + str(i) + "-" + str(j)
        os.mkdir(model_dir)
        
        def input_fn_train():
            return dict({k: tf.constant(sellers.iloc[train_indices][k].values, shape=[sellers.iloc[train_indices][k].size, 1])
                          for k in CONTINUOUS_COLUMNS}), tf.constant(sellers.iloc[train_indices][LABEL_COLUMN].values)

        def input_fn_eval():
            return dict({k: tf.constant(sellers.iloc[test_indices][k].values, shape=[sellers.iloc[test_indices][k].size, 1])
                          for k in CONTINUOUS_COLUMNS}), tf.constant(sellers.iloc[test_indices][LABEL_COLUMN].values)
        
        m=tf.contrib.learn.DNNRegressor(model_dir=model_dir,feature_columns=deep_columns, hidden_units=layer_sizes)
        
        m.fit(input_fn=input_fn_train , steps=step)
        time.sleep(1) 
        results = m.evaluate(input_fn=input_fn_eval, steps=1)
        del m 
        gc.collect()
        resultzz.append(results['loss'])

        j = j+1
        
    resultz.append(resultzz)