import os
os.environ['KERAS_BACKEND'] = 'torch'

import keras

import numpy as np
import pandas as pd

import fidle

# Init Fidle environment
run_id, run_dir, datasets_dir = fidle.init('K3WINE1')

Version              : 2.3.2
Run id               : K3WINE1
Run dir              : ./run/K3WINE1
Datasets dir         : /lustre/fswork/projects/rech/mlh/uja62cb/fidle-project/datasets-fidle
Start time           : 22/12/24 21:21:04
Hostname             : r3i7n1 (Linux)
Tensorflow log level : Info + Warning + Error  (=0)
Update keras cache   : False
Update torch cache   : False
Save figs            : ./run/K3WINE1/figs (True)
keras                : 3.7.0
numpy                : 2.1.2
sklearn              : 1.5.2
yaml                 : 6.0.2
matplotlib           : 3.9.2
pandas               : 2.2.3
torch                : 2.5.0

fit_verbosity = 1
dataset_name  = 'winequality-red.csv'

fidle.override('fit_verbosity', 'dataset_name')

** Overrided parameters : **
fit_verbosity        : 2

data = pd.read_csv(f'{datasets_dir}/WineQuality/origine/{dataset_name}', header=0,sep=';')

display(data.head(5).style.format("{0:.2f}"))
print('Missing Data : ',data.isna().sum().sum(), '  Shape is : ', data.shape)

Missing Data :  0   Shape is :  (1599, 12)

# ---- Split => train, test
#
data       = data.sample(frac=1., axis=0)     # Shuffle
data_train = data.sample(frac=0.8, axis=0)    # get 80 %
data_test  = data.drop(data_train.index)      # test = all - train

# ---- Split => x,y (medv is price)
#
x_train = data_train.drop('quality',  axis=1)
y_train = data_train['quality']
x_test  = data_test.drop('quality',   axis=1)
y_test  = data_test['quality']

print('Original data shape was : ',data.shape)
print('x_train : ',x_train.shape, 'y_train : ',y_train.shape)
print('x_test  : ',x_test.shape,  'y_test  : ',y_test.shape)

Original data shape was :  (1599, 12)
x_train :  (1279, 11) y_train :  (1279,)
x_test  :  (320, 11) y_test  :  (320,)

display(x_train.describe().style.format("{0:.2f}").set_caption("Before normalization :"))

mean = x_train.mean()
std  = x_train.std()
x_train = (x_train - mean) / std
x_test  = (x_test  - mean) / std

display(x_train.describe().style.format("{0:.2f}").set_caption("After normalization :"))

# Convert ou DataFrame to numpy array
x_train, y_train = np.array(x_train), np.array(y_train)
x_test,  y_test  = np.array(x_test),  np.array(y_test)

def get_model_v1(shape):
  
  model = keras.models.Sequential()
  model.add(keras.layers.Input(shape, name="InputLayer"))
  model.add(keras.layers.Dense(64, activation='relu', name='Dense_n1'))
  model.add(keras.layers.Dense(64, activation='relu', name='Dense_n2'))
  model.add(keras.layers.Dense(1, name='Output'))

  model.compile(optimizer = 'rmsprop',
                loss      = 'mse',
                metrics   = ['mae', 'mse'] )
  return model

model=get_model_v1( (11,) )

model.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓
┃ Layer (type)                         ┃ Output Shape                ┃         Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩
│ Dense_n1 (Dense)                     │ (None, 64)                  │             768 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ Dense_n2 (Dense)                     │ (None, 64)                  │           4,160 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ Output (Dense)                       │ (None, 1)                   │              65 │
└──────────────────────────────────────┴─────────────────────────────┴─────────────────┘

 Total params: 4,993 (19.50 KB)

 Trainable params: 4,993 (19.50 KB)

 Non-trainable params: 0 (0.00 B)

os.makedirs('./run/models',   mode=0o750, exist_ok=True)
save_dir = "./run/models/best_model.keras"

savemodel_callback = keras.callbacks.ModelCheckpoint( filepath=save_dir, monitor='val_mae', mode='max', save_best_only=True)

history = model.fit(x_train,
                    y_train,
                    epochs          = 100,
                    batch_size      = 10,
                    verbose         = fit_verbosity,
                    validation_data = (x_test, y_test),
                    callbacks       = [savemodel_callback])

Epoch 1/100

128/128 - 2s - 15ms/step - loss: 6.8173 - mae: 2.0394 - mse: 6.8173 - val_loss: 1.9679 - val_mae: 1.0980 - val_mse: 1.9679

Epoch 2/100

128/128 - 1s - 6ms/step - loss: 1.4374 - mae: 0.9350 - mse: 1.4374 - val_loss: 1.1838 - val_mae: 0.8539 - val_mse: 1.1838

Epoch 3/100

128/128 - 1s - 6ms/step - loss: 0.9545 - mae: 0.7510 - mse: 0.9545 - val_loss: 0.8727 - val_mae: 0.7226 - val_mse: 0.8727

Epoch 4/100

128/128 - 1s - 6ms/step - loss: 0.7503 - mae: 0.6711 - mse: 0.7503 - val_loss: 0.6673 - val_mae: 0.6248 - val_mse: 0.6673

Epoch 5/100

128/128 - 1s - 6ms/step - loss: 0.6203 - mae: 0.6117 - mse: 0.6203 - val_loss: 0.5975 - val_mae: 0.5857 - val_mse: 0.5975

score = model.evaluate(x_test, y_test, verbose=0)

print('x_test / loss      : {:5.4f}'.format(score[0]))
print('x_test / mae       : {:5.4f}'.format(score[1]))
print('x_test / mse       : {:5.4f}'.format(score[2]))

x_test / loss      : 0.3815
x_test / mae       : 0.4573
x_test / mse       : 0.3815

print("min( val_mae ) : {:.4f}".format( min(history.history["val_mae"]) ) )

min( val_mae ) : 0.4264

fidle.scrawler.history( history, plot={'MSE' :['mse', 'val_mse'],
                        'MAE' :['mae', 'val_mae'],
                        'LOSS':['loss','val_loss']}, save_as='01-history')

loaded_model = keras.models.load_model('./run/models/best_model.keras')
loaded_model.summary()
print("Loaded.")

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓
┃ Layer (type)                         ┃ Output Shape                ┃         Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩
│ Dense_n1 (Dense)                     │ (None, 64)                  │             768 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ Dense_n2 (Dense)                     │ (None, 64)                  │           4,160 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ Output (Dense)                       │ (None, 1)                   │              65 │
└──────────────────────────────────────┴─────────────────────────────┴─────────────────┘

 Total params: 9,988 (39.02 KB)

 Trainable params: 4,993 (19.50 KB)

 Non-trainable params: 0 (0.00 B)

 Optimizer params: 4,995 (19.51 KB)

Loaded.

score = loaded_model.evaluate(x_test, y_test, verbose=0)

print('x_test / loss      : {:5.4f}'.format(score[0]))
print('x_test / mae       : {:5.4f}'.format(score[1]))
print('x_test / mse       : {:5.4f}'.format(score[2]))

x_test / loss      : 1.9679
x_test / mae       : 1.0980
x_test / mse       : 1.9679

# ---- Pick n entries from our test set
n = 200
ii = np.random.randint(1,len(x_test),n)
x_sample = x_test[ii]
y_sample = y_test[ii]

# ---- Make a predictions
y_pred = loaded_model.predict( x_sample, verbose=2 )

7/7 - 0s - 2ms/step

# ---- Show it
print('Wine    Prediction   Real   Delta')
for i in range(n):
    pred   = y_pred[i][0]
    real   = y_sample[i]
    delta  = real-pred
    print(f'{i:03d}        {pred:.2f}       {real}      {delta:+.2f} ')

Wine    Prediction   Real   Delta
000        4.33       5      +0.67 
001        4.27       5      +0.73 
002        3.99       5      +1.01 
003        4.64       6      +1.36 
004        5.46       5      -0.46 
005        5.64       6      +0.36 
006        2.72       5      +2.28 
007        4.65       5      +0.35 
008        4.06       6      +1.94 
009        8.13       5      -3.13 
010        6.54       6      -0.54 
011        6.59       5      -1.59 
012        4.28       6      +1.72 
013        4.86       6      +1.14 
014        3.91       5      +1.09 
015        4.74       6      +1.26 
016        5.51       6      +0.49 
017        5.56       7      +1.44 
018        6.47       6      -0.47 
019        9.33       4      -5.33 
020        6.70       5      -1.70 
021        5.39       5      -0.39 
022        5.57       7      +1.43 
023        5.49       6      +0.51 
024        5.80       6      +0.20 
025        5.09       5      -0.09 
026        4.55       5      +0.45 
027        4.00       6      +2.00 
028        4.00       6      +2.00 
029        7.23       5      -2.23 
030        5.68       7      +1.32 
031        5.20       5      -0.20 
032        4.59       5      +0.41 
033        6.11       4      -2.11 
034        4.74       6      +1.26 
035        5.24       6      +0.76 
036        6.76       5      -1.76 
037        5.19       5      -0.19 
038        5.53       5      -0.53 
039        6.74       7      +0.26 
040        5.03       5      -0.03 
041        5.64       6      +0.36 
042        5.96       5      -0.96 
043        4.77       5      +0.23 
044        6.60       7      +0.40 
045        6.82       6      -0.82 
046        4.30       7      +2.70 
047        4.33       5      +0.67 
048        4.60       5      +0.40 
049        5.93       6      +0.07 
050        5.39       5      -0.39 
051        5.03       5      -0.03 
052        4.33       6      +1.67 
053        4.00       5      +1.00 
054        5.03       5      -0.03 
055        5.68       7      +1.32 
056        3.79       6      +2.21 
057        4.23       7      +2.77 
058        5.81       6      +0.19 
059        4.78       6      +1.22 
060        6.32       5      -1.32 
061        3.93       5      +1.07 
062        7.54       5      -2.54 
063        5.03       6      +0.97 
064        3.10       5      +1.90 
065        4.41       6      +1.59 
066        5.24       6      +0.76 
067        7.23       5      -2.23 
068        5.50       6      +0.50 
069        7.23       7      -0.23 
070        6.60       7      +0.40 
071        7.80       5      -2.80 
072        3.80       6      +2.20 
073        5.48       6      +0.52 
074        7.15       5      -2.15 
075        6.02       6      -0.02 
076        5.04       6      +0.96 
077        4.27       5      +0.73 
078        4.28       5      +0.72 
079        5.77       5      -0.77 
080        5.49       7      +1.51 
081        3.79       6      +2.21 
082        5.03       5      -0.03 
083        4.93       6      +1.07 
084        3.99       5      +1.01 
085        4.14       6      +1.86 
086        6.19       6      -0.19 
087        6.73       6      -0.73 
088        4.42       5      +0.58 
089        5.24       6      +0.76 
090        3.79       5      +1.21 
091        5.87       5      -0.87 
092        4.83       6      +1.17 
093        6.52       7      +0.48 
094        4.64       7      +2.36 
095        3.53       5      +1.47 
096        4.88       5      +0.12 
097        3.05       6      +2.95 
098        4.00       6      +2.00 
099        6.76       6      -0.76 
100        6.19       5      -1.19 
101        6.59       5      -1.59 
102        4.77       6      +1.23 
103        5.96       5      -0.96 
104        6.76       6      -0.76 
105        7.40       6      -1.40 
106        4.77       6      +1.23 
107        7.54       5      -2.54 
108        6.59       5      -1.59 
109        4.41       5      +0.59 
110        7.40       6      -1.40 
111        5.95       4      -1.95 
112        6.39       7      +0.61 
113        4.60       5      +0.40 
114        6.82       6      -0.82 
115        3.95       5      +1.05 
116        5.52       6      +0.48 
117        4.73       4      -0.73 
118        4.86       6      +1.14 
119        4.27       5      +0.73 
120        3.95       5      +1.05 
121        5.91       6      +0.09 
122        7.23       7      -0.23 
123        4.41       6      +1.59 
124        5.81       8      +2.19 
125        6.19       7      +0.81 
126        7.17       5      -2.17 
127        4.17       5      +0.83 
128        4.93       5      +0.07 
129        5.35       8      +2.65 
130        4.64       5      +0.36 
131        3.74       6      +2.26 
132        4.64       5      +0.36 
133        6.76       6      -0.76 
134        4.27       5      +0.73 
135        5.81       6      +0.19 
136        4.27       6      +1.73 
137        6.94       6      -0.94 
138        5.76       6      +0.24 
139        7.09       4      -3.09 
140        6.39       7      +0.61 
141        8.27       7      -1.27 
142        4.38       5      +0.62 
143        4.42       5      +0.58 
144        4.87       5      +0.13 
145        4.00       6      +2.00 
146        5.09       5      -0.09 
147        5.90       7      +1.10 
148        4.04       5      +0.96 
149        6.02       6      -0.02 
150        4.60       5      +0.40 
151        6.59       5      -1.59 
152        5.68       7      +1.32 
153        3.93       5      +1.07 
154        5.09       5      -0.09 
155        4.35       6      +1.65 
156        4.40       5      +0.60 
157        8.27       7      -1.27 
158        4.33       5      +0.67 
159        3.74       6      +2.26 
160        6.68       5      -1.68 
161        7.02       6      -1.02 
162        4.50       7      +2.50 
163        6.76       5      -1.76 
164        5.62       6      +0.38 
165        5.24       6      +0.76 
166        4.83       7      +2.17 
167        3.30       5      +1.70 
168        4.51       6      +1.49 
169        8.13       5      -3.13 
170        6.19       6      -0.19 
171        5.03       6      +0.97 
172        4.97       5      +0.03 
173        5.56       7      +1.44 
174        5.66       6      +0.34 
175        5.81       8      +2.19 
176        5.35       5      -0.35 
177        3.91       5      +1.09 
178        4.64       7      +2.36 
179        3.52       5      +1.48 
180        5.69       6      +0.31 
181        4.12       6      +1.88 
182        4.29       5      +0.71 
183        4.51       6      +1.49 
184        6.02       6      -0.02 
185        5.10       6      +0.90 
186        4.59       6      +1.41 
187        3.57       5      +1.43 
188        4.41       5      +0.59 
189        6.82       6      -0.82 
190        5.95       4      -1.95 
191        4.99       6      +1.01 
192        3.92       5      +1.08 
193        5.91       6      +0.09 
194        5.68       7      +1.32 
195        6.19       5      -1.19 
196        5.40       6      +0.60 
197        4.13       5      +0.87 
198        5.99       5      -0.99 
199        5.32       6      +0.68

fidle.end()

	fixed acidity	volatile acidity	citric acid	residual sugar	chlorides	free sulfur dioxide	total sulfur dioxide	density	pH	sulphates	alcohol	quality
0	7.40	0.70	0.00	1.90	0.08	11.00	34.00	1.00	3.51	0.56	9.40	5.00
1	7.80	0.88	0.00	2.60	0.10	25.00	67.00	1.00	3.20	0.68	9.80	5.00
2	7.80	0.76	0.04	2.30	0.09	15.00	54.00	1.00	3.26	0.65	9.80	5.00
3	11.20	0.28	0.56	1.90	0.07	17.00	60.00	1.00	3.16	0.58	9.80	6.00
4	7.40	0.70	0.00	1.90	0.08	11.00	34.00	1.00	3.51	0.56	9.40	5.00

	fixed acidity	volatile acidity	citric acid	residual sugar	chlorides	free sulfur dioxide	total sulfur dioxide	density	pH	sulphates	alcohol
count	1279.00	1279.00	1279.00	1279.00	1279.00	1279.00	1279.00	1279.00	1279.00	1279.00	1279.00
mean	8.32	0.53	0.27	2.53	0.09	15.99	46.53	1.00	3.31	0.66	10.41
std	1.75	0.18	0.20	1.40	0.04	10.37	32.83	0.00	0.15	0.17	1.06
min	4.60	0.12	0.00	0.90	0.01	1.00	6.00	0.99	2.74	0.33	8.40
25%	7.10	0.40	0.09	1.90	0.07	8.00	22.50	1.00	3.21	0.55	9.50
50%	7.90	0.52	0.25	2.20	0.08	14.00	38.00	1.00	3.31	0.62	10.10
75%	9.30	0.64	0.43	2.60	0.09	21.00	61.50	1.00	3.40	0.73	11.08
max	15.60	1.33	1.00	15.50	0.61	68.00	289.00	1.00	4.01	2.00	14.00

	fixed acidity	volatile acidity	citric acid	residual sugar	chlorides	free sulfur dioxide	total sulfur dioxide	density	pH	sulphates	alcohol
count	1279.00	1279.00	1279.00	1279.00	1279.00	1279.00	1279.00	1279.00	1279.00	1279.00	1279.00
mean	0.00	-0.00	-0.00	0.00	-0.00	0.00	-0.00	-0.00	-0.00	0.00	0.00
std	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
min	-2.13	-2.32	-1.37	-1.17	-1.70	-1.45	-1.23	-3.50	-3.72	-1.96	-1.89
25%	-0.70	-0.73	-0.91	-0.45	-0.38	-0.77	-0.73	-0.61	-0.66	-0.64	-0.85
50%	-0.24	-0.05	-0.09	-0.24	-0.18	-0.19	-0.26	0.02	-0.00	-0.23	-0.29
75%	0.56	0.62	0.83	0.05	0.07	0.48	0.46	0.56	0.58	0.43	0.64
max	4.16	4.53	3.74	9.27	11.89	5.02	7.39	3.63	4.56	8.01	3.38

[K3WINE1] - Wine quality prediction with a Dense Network (DNN)¶

Objectives :¶

What we're going to do :¶

Step 1 - Import and init¶

Step 2 - Retrieve data¶

Step 3 - Preparing the data¶

3.1 - Split data¶

3.2 - Data normalization¶

Step 4 - Build a model¶

5 - Train the model¶

5.1 - Get it¶

5.2 - Add callback¶

5.3 - Train it¶

Step 6 - Evaluate¶

6.1 - Model evaluation¶

6.2 - Training history¶

Step 7 - Restore a model :¶

7.1 - Reload model¶

7.2 - Evaluate it :¶

7.3 - Make a prediction¶

Few questions :¶