guide-multiple-outputs-with-keras-functional-api.ipynb

{"cells":[{"source":"<a href=\"https://www.kaggle.com/code/peremartramanonellas/guide-multiple-outputs-with-keras-functional-api?scriptVersionId=137650966\" target=\"_blank\"><img align=\"left\" alt=\"Kaggle\" title=\"Open in Kaggle\" src=\"https://kaggle.com/static/images/open-in-kaggle.svg\"></a>","metadata":{},"cell_type":"markdown"},{"cell_type":"code","execution_count":1,"id":"88a67528","metadata":{"_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","_kg_hide-input":true,"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","execution":{"iopub.execute_input":"2023-07-23T13:51:02.968463Z","iopub.status.busy":"2023-07-23T13:51:02.967758Z","iopub.status.idle":"2023-07-23T13:51:02.99097Z","shell.execute_reply":"2023-07-23T13:51:02.989791Z"},"papermill":{"duration":0.043147,"end_time":"2023-07-23T13:51:02.99459","exception":false,"start_time":"2023-07-23T13:51:02.951443","status":"completed"},"tags":[]},"outputs":[{"data":{"text/html":["\n","<style>\n",".output_png {\n","    display: table-cell;\n","    text-align: center;\n","    vertical-align: middle;\n","    horizontal-align: middle;\n","}\n","h1 {\n","    text-align: center;\n","    background-color: #6bacf5;\n","    padding: 10px;\n","    margin: 0;\n","    font-family: monospace;\n","    color:DimGray;\n","    border-radius: 2px\n","    style=\"font-family:verdana;\"\n","}\n","\n","h2 {\n","    text-align: center;\n","    background-color: #83c2ff;\n","    padding: 10px;\n","    margin: 0;\n","    font-family: monospace;\n","    color:DimGray;\n","    border-radius: 2px\n","}\n","\n","h3 {\n","    text-align: center;\n","    background-color: pink;\n","    padding: 10px;\n","    margin: 0;\n","    font-family: monospace;\n","    color:DimGray;\n","    border-radius: 2px\n","}\n","\n","h4 {\n","    text-align: center;\n","    background-color: pink;\n","    padding: 10px;\n","    margin: 0;\n","    font-family: monospace;\n","    color:DimGray;\n","    border-radius: 2px\n","}\n","\n","body, p {\n","    font-family: monospace;\n","    font-size: 18px;\n","    color: charcoal;\n","}\n","div {\n","    font-size: 14px;\n","    margin: 0;\n","\n","}\n","</style>\n"],"text/plain":["<IPython.core.display.HTML object>"]},"execution_count":1,"metadata":{},"output_type":"execute_result"}],"source":["from IPython.core.display import HTML\n","HTML(\"\"\"\n","<style>\n",".output_png {\n","    display: table-cell;\n","    text-align: center;\n","    vertical-align: middle;\n","    horizontal-align: middle;\n","}\n","h1 {\n","    text-align: center;\n","    background-color: #6bacf5;\n","    padding: 10px;\n","    margin: 0;\n","    font-family: monospace;\n","    color:DimGray;\n","    border-radius: 2px\n","    style=\"font-family:verdana;\"\n","}\n","\n","h2 {\n","    text-align: center;\n","    background-color: #83c2ff;\n","    padding: 10px;\n","    margin: 0;\n","    font-family: monospace;\n","    color:DimGray;\n","    border-radius: 2px\n","}\n","\n","h3 {\n","    text-align: center;\n","    background-color: pink;\n","    padding: 10px;\n","    margin: 0;\n","    font-family: monospace;\n","    color:DimGray;\n","    border-radius: 2px\n","}\n","\n","h4 {\n","    text-align: center;\n","    background-color: pink;\n","    padding: 10px;\n","    margin: 0;\n","    font-family: monospace;\n","    color:DimGray;\n","    border-radius: 2px\n","}\n","\n","body, p {\n","    font-family: monospace;\n","    font-size: 18px;\n","    color: charcoal;\n","}\n","div {\n","    font-size: 14px;\n","    margin: 0;\n","\n","}\n","</style>\n","\"\"\")"]},{"cell_type":"markdown","id":"bd964973","metadata":{"papermill":{"duration":0.01263,"end_time":"2023-07-23T13:51:03.020623","exception":false,"start_time":"2023-07-23T13:51:03.007993","status":"completed"},"tags":[]},"source":["# Tutorial to create a Multi Output Tensorflow Model using Keras API\n","\n","In this notebook we are going to create with TensorFlow a model able to predict two variables: \n","- **Classification variable** to identify the tipe of wine. \n","- **Regression variable** who qualifies the quality of the wine. \n","\n","I'm using a Dataset from Kaggle: https://www.kaggle.com/datasets/rajyellow46/wine-quality\n","It's a really simple tabular dataset containing some caracteristics of the wine, like acidity, sugar, chlorides... etc. And two values that can be labels: the type of wine and the quality. \n","\n","I'm going to do some Data cleaning, but it is not the focus of the Notebook. The focus is to show how we can use the Keras functional API to create a Model able to predict both variables. \n","\n","## Are this kind of models useful? \n","YES!  Maybe not in an environment like Kaggle, where we are competing to obtain the best score. But in the real world this kind of models can be really useful. \n","\n","Yo can save a lot of time in training and prediction, and a lot of computation hours. You can save time and money if you can use the same model to predict more than one variable. \n","\n","Just imagine a pipeline where the model must be retrained every day with millions of data, and with this data we can predict two or more variables. You don't need to train two or more models, you can use just one training process to get all the predictions you want. \n","\n","Amazing!!!! \n","\n","This models may be more difficult to adjust because we need to share the same hyperparameters for all the variables. But you can use different layers, and it should be enough. \n","\n","### Feel Free to fork or edit the noteboook for you own convenience. Please consider ***UPVOTING IT***. It helps others to discover the notebook, and it encourages me to continue publishing."]},{"cell_type":"code","execution_count":2,"id":"ccdd0df1","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:03.048614Z","iopub.status.busy":"2023-07-23T13:51:03.04729Z","iopub.status.idle":"2023-07-23T13:51:11.292809Z","shell.execute_reply":"2023-07-23T13:51:11.291573Z"},"papermill":{"duration":8.262979,"end_time":"2023-07-23T13:51:11.296394","exception":false,"start_time":"2023-07-23T13:51:03.033415","status":"completed"},"tags":[]},"outputs":[],"source":["#Importing the libraries. \n","\n","import tensorflow as tf\n","from tensorflow.keras.models import Model\n","from tensorflow.keras.layers import Dense, Input\n","\n","import pandas as pd\n","import matplotlib.pyplot as plt\n","import seaborn as sns \n","import numpy as np\n","\n","from tensorflow.python.keras.utils.vis_utils import plot_model\n","import pydot\n","\n","from scipy.stats import norm\n","from scipy import stats\n","\n","from sklearn.model_selection import train_test_split"]},{"cell_type":"markdown","id":"1028e544","metadata":{"papermill":{"duration":0.012266,"end_time":"2023-07-23T13:51:11.321881","exception":false,"start_time":"2023-07-23T13:51:11.309615","status":"completed"},"tags":[]},"source":["# Load and Prepare the DATA"]},{"cell_type":"code","execution_count":3,"id":"b5d69109","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:11.349942Z","iopub.status.busy":"2023-07-23T13:51:11.348655Z","iopub.status.idle":"2023-07-23T13:51:11.389113Z","shell.execute_reply":"2023-07-23T13:51:11.387954Z"},"papermill":{"duration":0.057735,"end_time":"2023-07-23T13:51:11.39238","exception":false,"start_time":"2023-07-23T13:51:11.334645","status":"completed"},"tags":[]},"outputs":[],"source":["#Load the file. You have an link to the dataset in the head of the notebook. \n","df = pd.read_csv('/kaggle/input/wine-quality/winequalityN.csv')"]},{"cell_type":"code","execution_count":4,"id":"5afdbb82","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:11.421104Z","iopub.status.busy":"2023-07-23T13:51:11.42069Z","iopub.status.idle":"2023-07-23T13:51:11.454351Z","shell.execute_reply":"2023-07-23T13:51:11.452968Z"},"papermill":{"duration":0.050367,"end_time":"2023-07-23T13:51:11.457152","exception":false,"start_time":"2023-07-23T13:51:11.406785","status":"completed"},"tags":[]},"outputs":[{"data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>type</th>\n","      <th>fixed acidity</th>\n","      <th>volatile acidity</th>\n","      <th>citric acid</th>\n","      <th>residual sugar</th>\n","      <th>chlorides</th>\n","      <th>free sulfur dioxide</th>\n","      <th>total sulfur dioxide</th>\n","      <th>density</th>\n","      <th>pH</th>\n","      <th>sulphates</th>\n","      <th>alcohol</th>\n","      <th>quality</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>white</td>\n","      <td>7.0</td>\n","      <td>0.27</td>\n","      <td>0.36</td>\n","      <td>20.7</td>\n","      <td>0.045</td>\n","      <td>45.0</td>\n","      <td>170.0</td>\n","      <td>1.0010</td>\n","      <td>3.00</td>\n","      <td>0.45</td>\n","      <td>8.8</td>\n","      <td>6</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>white</td>\n","      <td>6.3</td>\n","      <td>0.30</td>\n","      <td>0.34</td>\n","      <td>1.6</td>\n","      <td>0.049</td>\n","      <td>14.0</td>\n","      <td>132.0</td>\n","      <td>0.9940</td>\n","      <td>3.30</td>\n","      <td>0.49</td>\n","      <td>9.5</td>\n","      <td>6</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>white</td>\n","      <td>8.1</td>\n","      <td>0.28</td>\n","      <td>0.40</td>\n","      <td>6.9</td>\n","      <td>0.050</td>\n","      <td>30.0</td>\n","      <td>97.0</td>\n","      <td>0.9951</td>\n","      <td>3.26</td>\n","      <td>0.44</td>\n","      <td>10.1</td>\n","      <td>6</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>white</td>\n","      <td>7.2</td>\n","      <td>0.23</td>\n","      <td>0.32</td>\n","      <td>8.5</td>\n","      <td>0.058</td>\n","      <td>47.0</td>\n","      <td>186.0</td>\n","      <td>0.9956</td>\n","      <td>3.19</td>\n","      <td>0.40</td>\n","      <td>9.9</td>\n","      <td>6</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>white</td>\n","      <td>7.2</td>\n","      <td>0.23</td>\n","      <td>0.32</td>\n","      <td>8.5</td>\n","      <td>0.058</td>\n","      <td>47.0</td>\n","      <td>186.0</td>\n","      <td>0.9956</td>\n","      <td>3.19</td>\n","      <td>0.40</td>\n","      <td>9.9</td>\n","      <td>6</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["    type  fixed acidity  volatile acidity  citric acid  residual sugar  \\\n","0  white            7.0              0.27         0.36            20.7   \n","1  white            6.3              0.30         0.34             1.6   \n","2  white            8.1              0.28         0.40             6.9   \n","3  white            7.2              0.23         0.32             8.5   \n","4  white            7.2              0.23         0.32             8.5   \n","\n","   chlorides  free sulfur dioxide  total sulfur dioxide  density    pH  \\\n","0      0.045                 45.0                 170.0   1.0010  3.00   \n","1      0.049                 14.0                 132.0   0.9940  3.30   \n","2      0.050                 30.0                  97.0   0.9951  3.26   \n","3      0.058                 47.0                 186.0   0.9956  3.19   \n","4      0.058                 47.0                 186.0   0.9956  3.19   \n","\n","   sulphates  alcohol  quality  \n","0       0.45      8.8        6  \n","1       0.49      9.5        6  \n","2       0.44     10.1        6  \n","3       0.40      9.9        6  \n","4       0.40      9.9        6  "]},"execution_count":4,"metadata":{},"output_type":"execute_result"}],"source":["df.head()"]},{"cell_type":"markdown","id":"f0334ab9","metadata":{"papermill":{"duration":0.012742,"end_time":"2023-07-23T13:51:11.483126","exception":false,"start_time":"2023-07-23T13:51:11.470384","status":"completed"},"tags":[]},"source":["We need to convert the column type from text to numeric. "]},{"cell_type":"code","execution_count":5,"id":"6660d726","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:11.511033Z","iopub.status.busy":"2023-07-23T13:51:11.51062Z","iopub.status.idle":"2023-07-23T13:51:11.526166Z","shell.execute_reply":"2023-07-23T13:51:11.524879Z"},"papermill":{"duration":0.033064,"end_time":"2023-07-23T13:51:11.529213","exception":false,"start_time":"2023-07-23T13:51:11.496149","status":"completed"},"tags":[]},"outputs":[],"source":["#Converting string column to categorical numeric. \n","df['type'] = df['type'].apply(lambda x:0 if (x == 'white') else 1)"]},{"cell_type":"code","execution_count":6,"id":"486fb519","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:11.558353Z","iopub.status.busy":"2023-07-23T13:51:11.557474Z","iopub.status.idle":"2023-07-23T13:51:11.590797Z","shell.execute_reply":"2023-07-23T13:51:11.589335Z"},"papermill":{"duration":0.052734,"end_time":"2023-07-23T13:51:11.595427","exception":false,"start_time":"2023-07-23T13:51:11.542693","status":"completed"},"tags":[]},"outputs":[{"data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>type</th>\n","      <th>fixed acidity</th>\n","      <th>volatile acidity</th>\n","      <th>citric acid</th>\n","      <th>residual sugar</th>\n","      <th>chlorides</th>\n","      <th>free sulfur dioxide</th>\n","      <th>total sulfur dioxide</th>\n","      <th>density</th>\n","      <th>pH</th>\n","      <th>sulphates</th>\n","      <th>alcohol</th>\n","      <th>quality</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>0</td>\n","      <td>7.0</td>\n","      <td>0.270</td>\n","      <td>0.36</td>\n","      <td>20.7</td>\n","      <td>0.045</td>\n","      <td>45.0</td>\n","      <td>170.0</td>\n","      <td>1.00100</td>\n","      <td>3.00</td>\n","      <td>0.45</td>\n","      <td>8.8</td>\n","      <td>6</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>0</td>\n","      <td>6.3</td>\n","      <td>0.300</td>\n","      <td>0.34</td>\n","      <td>1.6</td>\n","      <td>0.049</td>\n","      <td>14.0</td>\n","      <td>132.0</td>\n","      <td>0.99400</td>\n","      <td>3.30</td>\n","      <td>0.49</td>\n","      <td>9.5</td>\n","      <td>6</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>0</td>\n","      <td>8.1</td>\n","      <td>0.280</td>\n","      <td>0.40</td>\n","      <td>6.9</td>\n","      <td>0.050</td>\n","      <td>30.0</td>\n","      <td>97.0</td>\n","      <td>0.99510</td>\n","      <td>3.26</td>\n","      <td>0.44</td>\n","      <td>10.1</td>\n","      <td>6</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>0</td>\n","      <td>7.2</td>\n","      <td>0.230</td>\n","      <td>0.32</td>\n","      <td>8.5</td>\n","      <td>0.058</td>\n","      <td>47.0</td>\n","      <td>186.0</td>\n","      <td>0.99560</td>\n","      <td>3.19</td>\n","      <td>0.40</td>\n","      <td>9.9</td>\n","      <td>6</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>0</td>\n","      <td>7.2</td>\n","      <td>0.230</td>\n","      <td>0.32</td>\n","      <td>8.5</td>\n","      <td>0.058</td>\n","      <td>47.0</td>\n","      <td>186.0</td>\n","      <td>0.99560</td>\n","      <td>3.19</td>\n","      <td>0.40</td>\n","      <td>9.9</td>\n","      <td>6</td>\n","    </tr>\n","    <tr>\n","      <th>...</th>\n","      <td>...</td>\n","      <td>...</td>\n","      <td>...</td>\n","      <td>...</td>\n","      <td>...</td>\n","      <td>...</td>\n","      <td>...</td>\n","      <td>...</td>\n","      <td>...</td>\n","      <td>...</td>\n","      <td>...</td>\n","      <td>...</td>\n","      <td>...</td>\n","    </tr>\n","    <tr>\n","      <th>6492</th>\n","      <td>1</td>\n","      <td>6.2</td>\n","      <td>0.600</td>\n","      <td>0.08</td>\n","      <td>2.0</td>\n","      <td>0.090</td>\n","      <td>32.0</td>\n","      <td>44.0</td>\n","      <td>0.99490</td>\n","      <td>3.45</td>\n","      <td>0.58</td>\n","      <td>10.5</td>\n","      <td>5</td>\n","    </tr>\n","    <tr>\n","      <th>6493</th>\n","      <td>1</td>\n","      <td>5.9</td>\n","      <td>0.550</td>\n","      <td>0.10</td>\n","      <td>2.2</td>\n","      <td>0.062</td>\n","      <td>39.0</td>\n","      <td>51.0</td>\n","      <td>0.99512</td>\n","      <td>3.52</td>\n","      <td>NaN</td>\n","      <td>11.2</td>\n","      <td>6</td>\n","    </tr>\n","    <tr>\n","      <th>6494</th>\n","      <td>1</td>\n","      <td>6.3</td>\n","      <td>0.510</td>\n","      <td>0.13</td>\n","      <td>2.3</td>\n","      <td>0.076</td>\n","      <td>29.0</td>\n","      <td>40.0</td>\n","      <td>0.99574</td>\n","      <td>3.42</td>\n","      <td>0.75</td>\n","      <td>11.0</td>\n","      <td>6</td>\n","    </tr>\n","    <tr>\n","      <th>6495</th>\n","      <td>1</td>\n","      <td>5.9</td>\n","      <td>0.645</td>\n","      <td>0.12</td>\n","      <td>2.0</td>\n","      <td>0.075</td>\n","      <td>32.0</td>\n","      <td>44.0</td>\n","      <td>0.99547</td>\n","      <td>3.57</td>\n","      <td>0.71</td>\n","      <td>10.2</td>\n","      <td>5</td>\n","    </tr>\n","    <tr>\n","      <th>6496</th>\n","      <td>1</td>\n","      <td>6.0</td>\n","      <td>0.310</td>\n","      <td>0.47</td>\n","      <td>3.6</td>\n","      <td>0.067</td>\n","      <td>18.0</td>\n","      <td>42.0</td>\n","      <td>0.99549</td>\n","      <td>3.39</td>\n","      <td>0.66</td>\n","      <td>11.0</td>\n","      <td>6</td>\n","    </tr>\n","  </tbody>\n","</table>\n","<p>6497 rows × 13 columns</p>\n","</div>"],"text/plain":["      type  fixed acidity  volatile acidity  citric acid  residual sugar  \\\n","0        0            7.0             0.270         0.36            20.7   \n","1        0            6.3             0.300         0.34             1.6   \n","2        0            8.1             0.280         0.40             6.9   \n","3        0            7.2             0.230         0.32             8.5   \n","4        0            7.2             0.230         0.32             8.5   \n","...    ...            ...               ...          ...             ...   \n","6492     1            6.2             0.600         0.08             2.0   \n","6493     1            5.9             0.550         0.10             2.2   \n","6494     1            6.3             0.510         0.13             2.3   \n","6495     1            5.9             0.645         0.12             2.0   \n","6496     1            6.0             0.310         0.47             3.6   \n","\n","      chlorides  free sulfur dioxide  total sulfur dioxide  density    pH  \\\n","0         0.045                 45.0                 170.0  1.00100  3.00   \n","1         0.049                 14.0                 132.0  0.99400  3.30   \n","2         0.050                 30.0                  97.0  0.99510  3.26   \n","3         0.058                 47.0                 186.0  0.99560  3.19   \n","4         0.058                 47.0                 186.0  0.99560  3.19   \n","...         ...                  ...                   ...      ...   ...   \n","6492      0.090                 32.0                  44.0  0.99490  3.45   \n","6493      0.062                 39.0                  51.0  0.99512  3.52   \n","6494      0.076                 29.0                  40.0  0.99574  3.42   \n","6495      0.075                 32.0                  44.0  0.99547  3.57   \n","6496      0.067                 18.0                  42.0  0.99549  3.39   \n","\n","      sulphates  alcohol  quality  \n","0          0.45      8.8        6  \n","1          0.49      9.5        6  \n","2          0.44     10.1        6  \n","3          0.40      9.9        6  \n","4          0.40      9.9        6  \n","...         ...      ...      ...  \n","6492       0.58     10.5        5  \n","6493        NaN     11.2        6  \n","6494       0.75     11.0        6  \n","6495       0.71     10.2        5  \n","6496       0.66     11.0        6  \n","\n","[6497 rows x 13 columns]"]},"execution_count":6,"metadata":{},"output_type":"execute_result"}],"source":["df"]},{"cell_type":"markdown","id":"2d4ecac7","metadata":{"papermill":{"duration":0.013096,"end_time":"2023-07-23T13:51:11.622165","exception":false,"start_time":"2023-07-23T13:51:11.609069","status":"completed"},"tags":[]},"source":["Our *type* colum now is númeric. \n","\n","Let's check if we have some null values. "]},{"cell_type":"code","execution_count":7,"id":"c39ec768","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:11.651932Z","iopub.status.busy":"2023-07-23T13:51:11.65103Z","iopub.status.idle":"2023-07-23T13:51:11.662648Z","shell.execute_reply":"2023-07-23T13:51:11.661377Z"},"papermill":{"duration":0.029837,"end_time":"2023-07-23T13:51:11.665632","exception":false,"start_time":"2023-07-23T13:51:11.635795","status":"completed"},"tags":[]},"outputs":[{"data":{"text/plain":["fixed acidity          10\n","pH                      9\n","volatile acidity        8\n","sulphates               4\n","citric acid             3\n","residual sugar          2\n","chlorides               2\n","type                    0\n","free sulfur dioxide     0\n","dtype: int64"]},"execution_count":7,"metadata":{},"output_type":"execute_result"}],"source":["#Get the numbers of Null values in columns, in descending order. \n","df.isna().sum().sort_values(ascending=False).head(9)"]},{"cell_type":"markdown","id":"55150e13","metadata":{"papermill":{"duration":0.013339,"end_time":"2023-07-23T13:51:11.692663","exception":false,"start_time":"2023-07-23T13:51:11.679324","status":"completed"},"tags":[]},"source":["There are just a few rows with null / empty values, we can just delete it. "]},{"cell_type":"code","execution_count":8,"id":"23cd24c3","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:11.721954Z","iopub.status.busy":"2023-07-23T13:51:11.72152Z","iopub.status.idle":"2023-07-23T13:51:11.735332Z","shell.execute_reply":"2023-07-23T13:51:11.734125Z"},"papermill":{"duration":0.031681,"end_time":"2023-07-23T13:51:11.738076","exception":false,"start_time":"2023-07-23T13:51:11.706395","status":"completed"},"tags":[]},"outputs":[],"source":["#Creating the new dataset without null, or empty, values. \n","df = df.dropna(axis=0)"]},{"cell_type":"code","execution_count":9,"id":"b17ef826","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:11.767155Z","iopub.status.busy":"2023-07-23T13:51:11.766737Z","iopub.status.idle":"2023-07-23T13:51:11.77752Z","shell.execute_reply":"2023-07-23T13:51:11.776545Z"},"papermill":{"duration":0.028065,"end_time":"2023-07-23T13:51:11.779858","exception":false,"start_time":"2023-07-23T13:51:11.751793","status":"completed"},"tags":[]},"outputs":[{"data":{"text/plain":["type                    0\n","fixed acidity           0\n","volatile acidity        0\n","citric acid             0\n","residual sugar          0\n","chlorides               0\n","free sulfur dioxide     0\n","total sulfur dioxide    0\n","dtype: int64"]},"execution_count":9,"metadata":{},"output_type":"execute_result"}],"source":["#Check if we have more nulls. \n","df.isna().sum().sort_values(ascending=False).head(8)"]},{"cell_type":"markdown","id":"534b931f","metadata":{"papermill":{"duration":0.013455,"end_time":"2023-07-23T13:51:11.80742","exception":false,"start_time":"2023-07-23T13:51:11.793965","status":"completed"},"tags":[]},"source":["Now we are going to create 3 Datasets. Train, Test and Validation Data. Using the function *train_test_split* from the *sklearn.models*. "]},{"cell_type":"code","execution_count":10,"id":"feecd788","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:11.83754Z","iopub.status.busy":"2023-07-23T13:51:11.83668Z","iopub.status.idle":"2023-07-23T13:51:11.847878Z","shell.execute_reply":"2023-07-23T13:51:11.846878Z"},"papermill":{"duration":0.029412,"end_time":"2023-07-23T13:51:11.850612","exception":false,"start_time":"2023-07-23T13:51:11.8212","status":"completed"},"tags":[]},"outputs":[],"source":["#Create the train and test data. 80% tain and 20 data. \n","train, test = train_test_split(df, test_size=0.2, random_state = 1)\n","\n","#From the train Data we are going to get a 20% more to create the validation data. \n","train, val = train_test_split(train, test_size=0.2, random_state = 1)"]},{"cell_type":"code","execution_count":11,"id":"e996cb59","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:11.880501Z","iopub.status.busy":"2023-07-23T13:51:11.88003Z","iopub.status.idle":"2023-07-23T13:51:11.886585Z","shell.execute_reply":"2023-07-23T13:51:11.88536Z"},"papermill":{"duration":0.024587,"end_time":"2023-07-23T13:51:11.889104","exception":false,"start_time":"2023-07-23T13:51:11.864517","status":"completed"},"tags":[]},"outputs":[],"source":["#With this function we got the labels *type* and *quality*,  ready to pass to the Model. \n","def get_labels(df):\n","    type_wine = df.pop('type')\n","    type_wine = np.array(type_wine)\n","    quality = df.pop('quality')\n","    quality = np.array(quality)\n","    return (quality, type_wine)"]},{"cell_type":"code","execution_count":12,"id":"d1708c55","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:11.918682Z","iopub.status.busy":"2023-07-23T13:51:11.918241Z","iopub.status.idle":"2023-07-23T13:51:11.927419Z","shell.execute_reply":"2023-07-23T13:51:11.926223Z"},"papermill":{"duration":0.027084,"end_time":"2023-07-23T13:51:11.930129","exception":false,"start_time":"2023-07-23T13:51:11.903045","status":"completed"},"tags":[]},"outputs":[],"source":["#Getting the labels for train, test and validate. \n","train_y = get_labels(train)\n","test_y = get_labels(test)\n","val_y = get_labels(val)"]},{"cell_type":"markdown","id":"2ace07b8","metadata":{"execution":{"iopub.execute_input":"2022-10-05T21:25:15.924654Z","iopub.status.busy":"2022-10-05T21:25:15.924242Z","iopub.status.idle":"2022-10-05T21:25:15.93019Z","shell.execute_reply":"2022-10-05T21:25:15.928874Z","shell.execute_reply.started":"2022-10-05T21:25:15.924619Z"},"papermill":{"duration":0.013423,"end_time":"2023-07-23T13:51:11.957847","exception":false,"start_time":"2023-07-23T13:51:11.944424","status":"completed"},"tags":[]},"source":["## Scale values"]},{"cell_type":"markdown","id":"30df7c4b","metadata":{"papermill":{"duration":0.013585,"end_time":"2023-07-23T13:51:11.985272","exception":false,"start_time":"2023-07-23T13:51:11.971687","status":"completed"},"tags":[]},"source":["This is the only treatment that I'm going to do to the Data. It's really simple, and improve a lot the result of the model. "]},{"cell_type":"code","execution_count":13,"id":"e7641c10","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:12.014842Z","iopub.status.busy":"2023-07-23T13:51:12.01438Z","iopub.status.idle":"2023-07-23T13:51:12.073522Z","shell.execute_reply":"2023-07-23T13:51:12.072446Z"},"papermill":{"duration":0.077255,"end_time":"2023-07-23T13:51:12.076321","exception":false,"start_time":"2023-07-23T13:51:11.999066","status":"completed"},"tags":[]},"outputs":[{"data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>count</th>\n","      <th>mean</th>\n","      <th>std</th>\n","      <th>min</th>\n","      <th>25%</th>\n","      <th>50%</th>\n","      <th>75%</th>\n","      <th>max</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>fixed acidity</th>\n","      <td>4136.0</td>\n","      <td>7.196905</td>\n","      <td>1.296497</td>\n","      <td>3.80000</td>\n","      <td>6.4000</td>\n","      <td>7.000000</td>\n","      <td>7.62500</td>\n","      <td>15.90000</td>\n","    </tr>\n","    <tr>\n","      <th>volatile acidity</th>\n","      <td>4136.0</td>\n","      <td>0.337378</td>\n","      <td>0.163193</td>\n","      <td>0.08000</td>\n","      <td>0.2200</td>\n","      <td>0.290000</td>\n","      <td>0.40000</td>\n","      <td>1.58000</td>\n","    </tr>\n","    <tr>\n","      <th>citric acid</th>\n","      <td>4136.0</td>\n","      <td>0.319108</td>\n","      <td>0.144535</td>\n","      <td>0.00000</td>\n","      <td>0.2500</td>\n","      <td>0.310000</td>\n","      <td>0.39000</td>\n","      <td>1.66000</td>\n","    </tr>\n","    <tr>\n","      <th>residual sugar</th>\n","      <td>4136.0</td>\n","      <td>5.510638</td>\n","      <td>4.812892</td>\n","      <td>0.60000</td>\n","      <td>1.8000</td>\n","      <td>3.200000</td>\n","      <td>8.20000</td>\n","      <td>65.80000</td>\n","    </tr>\n","    <tr>\n","      <th>chlorides</th>\n","      <td>4136.0</td>\n","      <td>0.055627</td>\n","      <td>0.035004</td>\n","      <td>0.01200</td>\n","      <td>0.0380</td>\n","      <td>0.047000</td>\n","      <td>0.06400</td>\n","      <td>0.61100</td>\n","    </tr>\n","    <tr>\n","      <th>free sulfur dioxide</th>\n","      <td>4136.0</td>\n","      <td>30.732834</td>\n","      <td>17.917700</td>\n","      <td>1.00000</td>\n","      <td>17.0000</td>\n","      <td>29.000000</td>\n","      <td>42.00000</td>\n","      <td>289.00000</td>\n","    </tr>\n","    <tr>\n","      <th>total sulfur dioxide</th>\n","      <td>4136.0</td>\n","      <td>116.689313</td>\n","      <td>56.244277</td>\n","      <td>7.00000</td>\n","      <td>80.0000</td>\n","      <td>119.000000</td>\n","      <td>156.00000</td>\n","      <td>440.00000</td>\n","    </tr>\n","    <tr>\n","      <th>density</th>\n","      <td>4136.0</td>\n","      <td>0.994708</td>\n","      <td>0.003034</td>\n","      <td>0.98713</td>\n","      <td>0.9923</td>\n","      <td>0.994885</td>\n","      <td>0.99704</td>\n","      <td>1.03898</td>\n","    </tr>\n","    <tr>\n","      <th>pH</th>\n","      <td>4136.0</td>\n","      <td>3.218873</td>\n","      <td>0.160356</td>\n","      <td>2.72000</td>\n","      <td>3.1100</td>\n","      <td>3.210000</td>\n","      <td>3.32000</td>\n","      <td>4.01000</td>\n","    </tr>\n","    <tr>\n","      <th>sulphates</th>\n","      <td>4136.0</td>\n","      <td>0.530771</td>\n","      <td>0.148288</td>\n","      <td>0.23000</td>\n","      <td>0.4300</td>\n","      <td>0.510000</td>\n","      <td>0.60000</td>\n","      <td>2.00000</td>\n","    </tr>\n","    <tr>\n","      <th>alcohol</th>\n","      <td>4136.0</td>\n","      <td>10.480809</td>\n","      <td>1.190722</td>\n","      <td>8.00000</td>\n","      <td>9.5000</td>\n","      <td>10.300000</td>\n","      <td>11.30000</td>\n","      <td>14.90000</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["                       count        mean        std      min      25%  \\\n","fixed acidity         4136.0    7.196905   1.296497  3.80000   6.4000   \n","volatile acidity      4136.0    0.337378   0.163193  0.08000   0.2200   \n","citric acid           4136.0    0.319108   0.144535  0.00000   0.2500   \n","residual sugar        4136.0    5.510638   4.812892  0.60000   1.8000   \n","chlorides             4136.0    0.055627   0.035004  0.01200   0.0380   \n","free sulfur dioxide   4136.0   30.732834  17.917700  1.00000  17.0000   \n","total sulfur dioxide  4136.0  116.689313  56.244277  7.00000  80.0000   \n","density               4136.0    0.994708   0.003034  0.98713   0.9923   \n","pH                    4136.0    3.218873   0.160356  2.72000   3.1100   \n","sulphates             4136.0    0.530771   0.148288  0.23000   0.4300   \n","alcohol               4136.0   10.480809   1.190722  8.00000   9.5000   \n","\n","                             50%        75%        max  \n","fixed acidity           7.000000    7.62500   15.90000  \n","volatile acidity        0.290000    0.40000    1.58000  \n","citric acid             0.310000    0.39000    1.66000  \n","residual sugar          3.200000    8.20000   65.80000  \n","chlorides               0.047000    0.06400    0.61100  \n","free sulfur dioxide    29.000000   42.00000  289.00000  \n","total sulfur dioxide  119.000000  156.00000  440.00000  \n","density                 0.994885    0.99704    1.03898  \n","pH                      3.210000    3.32000    4.01000  \n","sulphates               0.510000    0.60000    2.00000  \n","alcohol                10.300000   11.30000   14.90000  "]},"execution_count":13,"metadata":{},"output_type":"execute_result"}],"source":["#Get Mean and std but only from training data. \n","train_stats = train.describe()\n","train_stats = train_stats.transpose()\n","train_stats"]},{"cell_type":"markdown","id":"62a991fa","metadata":{"papermill":{"duration":0.013895,"end_time":"2023-07-23T13:51:12.104772","exception":false,"start_time":"2023-07-23T13:51:12.090877","status":"completed"},"tags":[]},"source":["> Note that I used only the train dataset to calculate the mean and std. **It's a good practice don't use the validation or test data to calculate this values** because is how it's going to be in the real world. The model only must know the data used for training, and therefore we must only this dataset to calculate the *mean* and *std*.\n","\n","I'm going to use the data contained in the *train_stats* to normalize the values in the datasets. "]},{"cell_type":"code","execution_count":14,"id":"e0256f65","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:12.135549Z","iopub.status.busy":"2023-07-23T13:51:12.13484Z","iopub.status.idle":"2023-07-23T13:51:12.140537Z","shell.execute_reply":"2023-07-23T13:51:12.139425Z"},"papermill":{"duration":0.02434,"end_time":"2023-07-23T13:51:12.143401","exception":false,"start_time":"2023-07-23T13:51:12.119061","status":"completed"},"tags":[]},"outputs":[],"source":["# Normalize the data, but with the mean and std of only train data. \n","def scale_data(df):\n","    return (df - train_stats['mean']) / train_stats['std']"]},{"cell_type":"code","execution_count":15,"id":"2de00913","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:12.174226Z","iopub.status.busy":"2023-07-23T13:51:12.173788Z","iopub.status.idle":"2023-07-23T13:51:12.183546Z","shell.execute_reply":"2023-07-23T13:51:12.18246Z"},"papermill":{"duration":0.028438,"end_time":"2023-07-23T13:51:12.186375","exception":false,"start_time":"2023-07-23T13:51:12.157937","status":"completed"},"tags":[]},"outputs":[],"source":["#Scaling the 3 datasets. \n","train_X = scale_data(train)\n","test_X = scale_data(test)\n","val_X = scale_data(val)"]},{"cell_type":"markdown","id":"8e1a7262","metadata":{"papermill":{"duration":0.014355,"end_time":"2023-07-23T13:51:12.215112","exception":false,"start_time":"2023-07-23T13:51:12.200757","status":"completed"},"tags":[]},"source":["# The Model"]},{"cell_type":"markdown","id":"aa2a5125","metadata":{"papermill":{"duration":0.013667,"end_time":"2023-07-23T13:51:12.243459","exception":false,"start_time":"2023-07-23T13:51:12.229792","status":"completed"},"tags":[]},"source":["We are going to use the Keras functional API to create the model. The main advantatge is that we don't need to create sequential models. \n","\n","In this Model we are going to follow 2 different branches with some base layers shared, but with different output layers and one Dense layer more in one of the branches. "]},{"attachments":{"98fb5767-ec71-4568-8092-05c260afd66c.png":{"image/png":"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"}},"cell_type":"markdown","id":"622a1789","metadata":{"papermill":{"duration":0.013699,"end_time":"2023-07-23T13:51:12.271277","exception":false,"start_time":"2023-07-23T13:51:12.257578","status":"completed"},"tags":[]},"source":["![modelplot.png](attachment:98fb5767-ec71-4568-8092-05c260afd66c.png)"]},{"cell_type":"markdown","id":"44cc0ea6","metadata":{"papermill":{"duration":0.013583,"end_time":"2023-07-23T13:51:12.299015","exception":false,"start_time":"2023-07-23T13:51:12.285432","status":"completed"},"tags":[]},"source":["As you can see we have an additionl *Dense layer* with 64 Neurons to calculate the wine quality that it's not in the branch to calculate the wine type. "]},{"cell_type":"code","execution_count":16,"id":"c464c64d","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:12.329841Z","iopub.status.busy":"2023-07-23T13:51:12.329381Z","iopub.status.idle":"2023-07-23T13:51:12.485306Z","shell.execute_reply":"2023-07-23T13:51:12.484121Z"},"papermill":{"duration":0.174469,"end_time":"2023-07-23T13:51:12.488252","exception":false,"start_time":"2023-07-23T13:51:12.313783","status":"completed"},"tags":[]},"outputs":[],"source":["#Start with the input layer, where we must indicate the shape of the Data passed to the model. \n","inputs = tf.keras.layers.Input(shape=(11,))\n","\n","#Add dense layers to the input layer. These layers are commom to both predicted variables. \n","x = Dense(units=32, activation='relu')(inputs)\n","x = Dense(units=32, activation='relu')(x)\n","\n","#Add the output layer for the Wine type using Sigmoid activation. \n","y_t_layer = Dense(units = 1, activation='sigmoid', name='y_t_layer')(x)\n","\n","#Here we diversificate the model adding a new Dense layer to the Base layers (x)\n","quality_layer=Dense(units=64, name='quality_layer', activation='relu')(x)\n","\n","#The output layer for the quality wine variable. It's added below the Dense Layer: quality_layer \n","y_q_layer = Dense(units=1, name='y_q_layer')(quality_layer)\n","\n","#The Model is created indicating the inputs and outputs. \n","#We have only one Input, but we can create models with multiple inputs. \n","#The name in outputs is the same of the variables, and the internal name of the layer. \n","model = Model(inputs=inputs, outputs=[y_q_layer, y_t_layer])\n","\n","#I tested two optimizers and choosed Adam, but feel free to test yourself. \n","optimizer = tf.keras.optimizers.RMSprop(learning_rate=0.0001)\n","#optimizer = tf.keras.optimizers.Adam()\n","\n","#To compile the model we use two dictionaries, to indicate the loss functions and metrics \n","#for each output layer. Note that the name of the layer must be the same than the \n","#internal name of the layer. \n","model.compile(optimizer=optimizer, \n","              loss = {'y_t_layer' : 'binary_crossentropy', \n","                      'y_q_layer' : 'mse'\n","                     },\n","              metrics = {'y_t_layer' : 'accuracy', \n","                         'y_q_layer': tf.keras.metrics.RootMeanSquaredError()\n","                       }\n","             )"]},{"cell_type":"code","execution_count":17,"id":"2ca2630f","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:12.518632Z","iopub.status.busy":"2023-07-23T13:51:12.518214Z","iopub.status.idle":"2023-07-23T13:51:12.526333Z","shell.execute_reply":"2023-07-23T13:51:12.525144Z"},"papermill":{"duration":0.026426,"end_time":"2023-07-23T13:51:12.529149","exception":false,"start_time":"2023-07-23T13:51:12.502723","status":"completed"},"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["Model: \"model\"\n","__________________________________________________________________________________________________\n","Layer (type)                    Output Shape         Param #     Connected to                     \n","==================================================================================================\n","input_1 (InputLayer)            [(None, 11)]         0                                            \n","__________________________________________________________________________________________________\n","dense (Dense)                   (None, 32)           384         input_1[0][0]                    \n","__________________________________________________________________________________________________\n","dense_1 (Dense)                 (None, 32)           1056        dense[0][0]                      \n","__________________________________________________________________________________________________\n","quality_layer (Dense)           (None, 64)           2112        dense_1[0][0]                    \n","__________________________________________________________________________________________________\n","y_q_layer (Dense)               (None, 1)            65          quality_layer[0][0]              \n","__________________________________________________________________________________________________\n","y_t_layer (Dense)               (None, 1)            33          dense_1[0][0]                    \n","==================================================================================================\n","Total params: 3,650\n","Trainable params: 3,650\n","Non-trainable params: 0\n","__________________________________________________________________________________________________\n"]}],"source":["model.summary()"]},{"cell_type":"code","execution_count":18,"id":"c56916e9","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:12.55965Z","iopub.status.busy":"2023-07-23T13:51:12.559232Z","iopub.status.idle":"2023-07-23T13:51:13.743772Z","shell.execute_reply":"2023-07-23T13:51:13.742289Z"},"papermill":{"duration":1.202949,"end_time":"2023-07-23T13:51:13.746521","exception":false,"start_time":"2023-07-23T13:51:12.543572","status":"completed"},"tags":[]},"outputs":[{"data":{"image/png":"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\n","text/plain":["<IPython.core.display.Image object>"]},"execution_count":18,"metadata":{},"output_type":"execute_result"}],"source":["tf.keras.utils.plot_model(model, show_shapes=True, show_layer_names=True)"]},{"cell_type":"code","execution_count":19,"id":"2295ee41","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:13.780545Z","iopub.status.busy":"2023-07-23T13:51:13.780064Z","iopub.status.idle":"2023-07-23T13:51:32.607754Z","shell.execute_reply":"2023-07-23T13:51:32.606684Z"},"papermill":{"duration":18.847777,"end_time":"2023-07-23T13:51:32.610616","exception":false,"start_time":"2023-07-23T13:51:13.762839","status":"completed"},"scrolled":true,"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["Epoch 1/40\n","130/130 [==============================] - 2s 6ms/step - loss: 35.1024 - y_q_layer_loss: 34.3714 - y_t_layer_loss: 0.7310 - y_q_layer_root_mean_squared_error: 5.8627 - y_t_layer_accuracy: 0.3936 - val_loss: 31.3185 - val_y_q_layer_loss: 30.5905 - val_y_t_layer_loss: 0.7280 - val_y_q_layer_root_mean_squared_error: 5.5309 - val_y_t_layer_accuracy: 0.4149\n","Epoch 2/40\n","130/130 [==============================] - 0s 3ms/step - loss: 27.8288 - y_q_layer_loss: 27.1091 - y_t_layer_loss: 0.7197 - y_q_layer_root_mean_squared_error: 5.2066 - y_t_layer_accuracy: 0.4456 - val_loss: 23.7766 - val_y_q_layer_loss: 23.0689 - val_y_t_layer_loss: 0.7077 - val_y_q_layer_root_mean_squared_error: 4.8030 - val_y_t_layer_accuracy: 0.5010\n","Epoch 3/40\n","130/130 [==============================] - 0s 3ms/step - loss: 19.6896 - y_q_layer_loss: 19.0010 - y_t_layer_loss: 0.6886 - y_q_layer_root_mean_squared_error: 4.3590 - y_t_layer_accuracy: 0.5532 - val_loss: 14.9144 - val_y_q_layer_loss: 14.2422 - val_y_t_layer_loss: 0.6722 - val_y_q_layer_root_mean_squared_error: 3.7739 - val_y_t_layer_accuracy: 0.5783\n","Epoch 4/40\n","130/130 [==============================] - 0s 3ms/step - loss: 11.0358 - y_q_layer_loss: 10.3939 - y_t_layer_loss: 0.6419 - y_q_layer_root_mean_squared_error: 3.2240 - y_t_layer_accuracy: 0.6489 - val_loss: 7.1997 - val_y_q_layer_loss: 6.5799 - val_y_t_layer_loss: 0.6198 - val_y_q_layer_root_mean_squared_error: 2.5651 - val_y_t_layer_accuracy: 0.6576\n","Epoch 5/40\n","130/130 [==============================] - 0s 3ms/step - loss: 5.4871 - y_q_layer_loss: 4.9085 - y_t_layer_loss: 0.5786 - y_q_layer_root_mean_squared_error: 2.2155 - y_t_layer_accuracy: 0.7309 - val_loss: 4.3772 - val_y_q_layer_loss: 3.8188 - val_y_t_layer_loss: 0.5584 - val_y_q_layer_root_mean_squared_error: 1.9542 - val_y_t_layer_accuracy: 0.7456\n","Epoch 6/40\n","130/130 [==============================] - 0s 4ms/step - loss: 3.9761 - y_q_layer_loss: 3.4627 - y_t_layer_loss: 0.5134 - y_q_layer_root_mean_squared_error: 1.8608 - y_t_layer_accuracy: 0.7868 - val_loss: 3.5571 - val_y_q_layer_loss: 3.0568 - val_y_t_layer_loss: 0.5003 - val_y_q_layer_root_mean_squared_error: 1.7484 - val_y_t_layer_accuracy: 0.7843\n","Epoch 7/40\n","130/130 [==============================] - 0s 3ms/step - loss: 3.3262 - y_q_layer_loss: 2.8701 - y_t_layer_loss: 0.4560 - y_q_layer_root_mean_squared_error: 1.6941 - y_t_layer_accuracy: 0.8187 - val_loss: 2.9919 - val_y_q_layer_loss: 2.5443 - val_y_t_layer_loss: 0.4476 - val_y_q_layer_root_mean_squared_error: 1.5951 - val_y_t_layer_accuracy: 0.8133\n","Epoch 8/40\n","130/130 [==============================] - 0s 3ms/step - loss: 2.8698 - y_q_layer_loss: 2.4628 - y_t_layer_loss: 0.4070 - y_q_layer_root_mean_squared_error: 1.5693 - y_t_layer_accuracy: 0.8373 - val_loss: 2.6011 - val_y_q_layer_loss: 2.1979 - val_y_t_layer_loss: 0.4032 - val_y_q_layer_root_mean_squared_error: 1.4825 - val_y_t_layer_accuracy: 0.8395\n","Epoch 9/40\n","130/130 [==============================] - 0s 3ms/step - loss: 2.5161 - y_q_layer_loss: 2.1524 - y_t_layer_loss: 0.3636 - y_q_layer_root_mean_squared_error: 1.4671 - y_t_layer_accuracy: 0.8757 - val_loss: 2.2736 - val_y_q_layer_loss: 1.9108 - val_y_t_layer_loss: 0.3627 - val_y_q_layer_root_mean_squared_error: 1.3823 - val_y_t_layer_accuracy: 0.8762\n","Epoch 10/40\n","130/130 [==============================] - 0s 3ms/step - loss: 2.2245 - y_q_layer_loss: 1.8989 - y_t_layer_loss: 0.3256 - y_q_layer_root_mean_squared_error: 1.3780 - y_t_layer_accuracy: 0.9122 - val_loss: 2.0179 - val_y_q_layer_loss: 1.6930 - val_y_t_layer_loss: 0.3249 - val_y_q_layer_root_mean_squared_error: 1.3011 - val_y_t_layer_accuracy: 0.9168\n","Epoch 11/40\n","130/130 [==============================] - 0s 3ms/step - loss: 1.9849 - y_q_layer_loss: 1.6954 - y_t_layer_loss: 0.2895 - y_q_layer_root_mean_squared_error: 1.3021 - y_t_layer_accuracy: 0.9410 - val_loss: 1.8125 - val_y_q_layer_loss: 1.5218 - val_y_t_layer_loss: 0.2907 - val_y_q_layer_root_mean_squared_error: 1.2336 - val_y_t_layer_accuracy: 0.9439\n","Epoch 12/40\n","130/130 [==============================] - 0s 3ms/step - loss: 1.8016 - y_q_layer_loss: 1.5437 - y_t_layer_loss: 0.2579 - y_q_layer_root_mean_squared_error: 1.2425 - y_t_layer_accuracy: 0.9553 - val_loss: 1.6543 - val_y_q_layer_loss: 1.3944 - val_y_t_layer_loss: 0.2599 - val_y_q_layer_root_mean_squared_error: 1.1808 - val_y_t_layer_accuracy: 0.9545\n","Epoch 13/40\n","130/130 [==============================] - 0s 3ms/step - loss: 1.6551 - y_q_layer_loss: 1.4248 - y_t_layer_loss: 0.2303 - y_q_layer_root_mean_squared_error: 1.1937 - y_t_layer_accuracy: 0.9630 - val_loss: 1.5176 - val_y_q_layer_loss: 1.2848 - val_y_t_layer_loss: 0.2328 - val_y_q_layer_root_mean_squared_error: 1.1335 - val_y_t_layer_accuracy: 0.9545\n","Epoch 14/40\n","130/130 [==============================] - 0s 3ms/step - loss: 1.5362 - y_q_layer_loss: 1.3300 - y_t_layer_loss: 0.2062 - y_q_layer_root_mean_squared_error: 1.1533 - y_t_layer_accuracy: 0.9698 - val_loss: 1.4081 - val_y_q_layer_loss: 1.1984 - val_y_t_layer_loss: 0.2098 - val_y_q_layer_root_mean_squared_error: 1.0947 - val_y_t_layer_accuracy: 0.9584\n","Epoch 15/40\n","130/130 [==============================] - 0s 3ms/step - loss: 1.4327 - y_q_layer_loss: 1.2468 - y_t_layer_loss: 0.1859 - y_q_layer_root_mean_squared_error: 1.1166 - y_t_layer_accuracy: 0.9744 - val_loss: 1.3135 - val_y_q_layer_loss: 1.1242 - val_y_t_layer_loss: 0.1893 - val_y_q_layer_root_mean_squared_error: 1.0603 - val_y_t_layer_accuracy: 0.9623\n","Epoch 16/40\n","130/130 [==============================] - 0s 3ms/step - loss: 1.3361 - y_q_layer_loss: 1.1689 - y_t_layer_loss: 0.1672 - y_q_layer_root_mean_squared_error: 1.0812 - y_t_layer_accuracy: 0.9765 - val_loss: 1.2306 - val_y_q_layer_loss: 1.0586 - val_y_t_layer_loss: 0.1721 - val_y_q_layer_root_mean_squared_error: 1.0289 - val_y_t_layer_accuracy: 0.9662\n","Epoch 17/40\n","130/130 [==============================] - 0s 3ms/step - loss: 1.2564 - y_q_layer_loss: 1.1046 - y_t_layer_loss: 0.1518 - y_q_layer_root_mean_squared_error: 1.0510 - y_t_layer_accuracy: 0.9778 - val_loss: 1.1664 - val_y_q_layer_loss: 1.0089 - val_y_t_layer_loss: 0.1574 - val_y_q_layer_root_mean_squared_error: 1.0045 - val_y_t_layer_accuracy: 0.9700\n","Epoch 18/40\n","130/130 [==============================] - 0s 3ms/step - loss: 1.1878 - y_q_layer_loss: 1.0491 - y_t_layer_loss: 0.1387 - y_q_layer_root_mean_squared_error: 1.0243 - y_t_layer_accuracy: 0.9807 - val_loss: 1.0980 - val_y_q_layer_loss: 0.9537 - val_y_t_layer_loss: 0.1443 - val_y_q_layer_root_mean_squared_error: 0.9766 - val_y_t_layer_accuracy: 0.9739\n","Epoch 19/40\n","130/130 [==============================] - 0s 3ms/step - loss: 1.1248 - y_q_layer_loss: 0.9976 - y_t_layer_loss: 0.1273 - y_q_layer_root_mean_squared_error: 0.9988 - y_t_layer_accuracy: 0.9831 - val_loss: 1.0440 - val_y_q_layer_loss: 0.9107 - val_y_t_layer_loss: 0.1333 - val_y_q_layer_root_mean_squared_error: 0.9543 - val_y_t_layer_accuracy: 0.9758\n","Epoch 20/40\n","130/130 [==============================] - 0s 3ms/step - loss: 1.0737 - y_q_layer_loss: 0.9562 - y_t_layer_loss: 0.1176 - y_q_layer_root_mean_squared_error: 0.9778 - y_t_layer_accuracy: 0.9843 - val_loss: 0.9998 - val_y_q_layer_loss: 0.8762 - val_y_t_layer_loss: 0.1235 - val_y_q_layer_root_mean_squared_error: 0.9361 - val_y_t_layer_accuracy: 0.9778\n","Epoch 21/40\n","130/130 [==============================] - 0s 3ms/step - loss: 1.0229 - y_q_layer_loss: 0.9139 - y_t_layer_loss: 0.1090 - y_q_layer_root_mean_squared_error: 0.9560 - y_t_layer_accuracy: 0.9857 - val_loss: 0.9520 - val_y_q_layer_loss: 0.8370 - val_y_t_layer_loss: 0.1150 - val_y_q_layer_root_mean_squared_error: 0.9149 - val_y_t_layer_accuracy: 0.9807\n","Epoch 22/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.9812 - y_q_layer_loss: 0.8795 - y_t_layer_loss: 0.1017 - y_q_layer_root_mean_squared_error: 0.9378 - y_t_layer_accuracy: 0.9865 - val_loss: 0.9152 - val_y_q_layer_loss: 0.8075 - val_y_t_layer_loss: 0.1077 - val_y_q_layer_root_mean_squared_error: 0.8986 - val_y_t_layer_accuracy: 0.9855\n","Epoch 23/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.9423 - y_q_layer_loss: 0.8470 - y_t_layer_loss: 0.0953 - y_q_layer_root_mean_squared_error: 0.9203 - y_t_layer_accuracy: 0.9865 - val_loss: 0.8920 - val_y_q_layer_loss: 0.7910 - val_y_t_layer_loss: 0.1010 - val_y_q_layer_root_mean_squared_error: 0.8894 - val_y_t_layer_accuracy: 0.9845\n","Epoch 24/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.9090 - y_q_layer_loss: 0.8196 - y_t_layer_loss: 0.0894 - y_q_layer_root_mean_squared_error: 0.9053 - y_t_layer_accuracy: 0.9865 - val_loss: 0.8524 - val_y_q_layer_loss: 0.7574 - val_y_t_layer_loss: 0.0950 - val_y_q_layer_root_mean_squared_error: 0.8703 - val_y_t_layer_accuracy: 0.9845\n","Epoch 25/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.8747 - y_q_layer_loss: 0.7904 - y_t_layer_loss: 0.0843 - y_q_layer_root_mean_squared_error: 0.8891 - y_t_layer_accuracy: 0.9879 - val_loss: 0.8342 - val_y_q_layer_loss: 0.7443 - val_y_t_layer_loss: 0.0899 - val_y_q_layer_root_mean_squared_error: 0.8627 - val_y_t_layer_accuracy: 0.9845\n","Epoch 26/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.8476 - y_q_layer_loss: 0.7677 - y_t_layer_loss: 0.0798 - y_q_layer_root_mean_squared_error: 0.8762 - y_t_layer_accuracy: 0.9882 - val_loss: 0.8020 - val_y_q_layer_loss: 0.7169 - val_y_t_layer_loss: 0.0851 - val_y_q_layer_root_mean_squared_error: 0.8467 - val_y_t_layer_accuracy: 0.9845\n","Epoch 27/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.8210 - y_q_layer_loss: 0.7454 - y_t_layer_loss: 0.0756 - y_q_layer_root_mean_squared_error: 0.8634 - y_t_layer_accuracy: 0.9886 - val_loss: 0.7779 - val_y_q_layer_loss: 0.6970 - val_y_t_layer_loss: 0.0809 - val_y_q_layer_root_mean_squared_error: 0.8349 - val_y_t_layer_accuracy: 0.9845\n","Epoch 28/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.7958 - y_q_layer_loss: 0.7237 - y_t_layer_loss: 0.0720 - y_q_layer_root_mean_squared_error: 0.8507 - y_t_layer_accuracy: 0.9889 - val_loss: 0.7619 - val_y_q_layer_loss: 0.6846 - val_y_t_layer_loss: 0.0773 - val_y_q_layer_root_mean_squared_error: 0.8274 - val_y_t_layer_accuracy: 0.9855\n","Epoch 29/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.7731 - y_q_layer_loss: 0.7043 - y_t_layer_loss: 0.0688 - y_q_layer_root_mean_squared_error: 0.8392 - y_t_layer_accuracy: 0.9889 - val_loss: 0.7371 - val_y_q_layer_loss: 0.6634 - val_y_t_layer_loss: 0.0738 - val_y_q_layer_root_mean_squared_error: 0.8145 - val_y_t_layer_accuracy: 0.9865\n","Epoch 30/40\n","130/130 [==============================] - 0s 4ms/step - loss: 0.7510 - y_q_layer_loss: 0.6854 - y_t_layer_loss: 0.0657 - y_q_layer_root_mean_squared_error: 0.8279 - y_t_layer_accuracy: 0.9894 - val_loss: 0.7205 - val_y_q_layer_loss: 0.6498 - val_y_t_layer_loss: 0.0707 - val_y_q_layer_root_mean_squared_error: 0.8061 - val_y_t_layer_accuracy: 0.9874\n","Epoch 31/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.7322 - y_q_layer_loss: 0.6692 - y_t_layer_loss: 0.0631 - y_q_layer_root_mean_squared_error: 0.8180 - y_t_layer_accuracy: 0.9896 - val_loss: 0.7070 - val_y_q_layer_loss: 0.6390 - val_y_t_layer_loss: 0.0680 - val_y_q_layer_root_mean_squared_error: 0.7994 - val_y_t_layer_accuracy: 0.9874\n","Epoch 32/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.7153 - y_q_layer_loss: 0.6546 - y_t_layer_loss: 0.0607 - y_q_layer_root_mean_squared_error: 0.8091 - y_t_layer_accuracy: 0.9894 - val_loss: 0.6924 - val_y_q_layer_loss: 0.6268 - val_y_t_layer_loss: 0.0656 - val_y_q_layer_root_mean_squared_error: 0.7917 - val_y_t_layer_accuracy: 0.9874\n","Epoch 33/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.6992 - y_q_layer_loss: 0.6406 - y_t_layer_loss: 0.0586 - y_q_layer_root_mean_squared_error: 0.8003 - y_t_layer_accuracy: 0.9901 - val_loss: 0.6804 - val_y_q_layer_loss: 0.6170 - val_y_t_layer_loss: 0.0634 - val_y_q_layer_root_mean_squared_error: 0.7855 - val_y_t_layer_accuracy: 0.9874\n","Epoch 34/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.6841 - y_q_layer_loss: 0.6274 - y_t_layer_loss: 0.0567 - y_q_layer_root_mean_squared_error: 0.7921 - y_t_layer_accuracy: 0.9903 - val_loss: 0.6692 - val_y_q_layer_loss: 0.6077 - val_y_t_layer_loss: 0.0614 - val_y_q_layer_root_mean_squared_error: 0.7796 - val_y_t_layer_accuracy: 0.9874\n","Epoch 35/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.6712 - y_q_layer_loss: 0.6163 - y_t_layer_loss: 0.0549 - y_q_layer_root_mean_squared_error: 0.7850 - y_t_layer_accuracy: 0.9906 - val_loss: 0.6592 - val_y_q_layer_loss: 0.5995 - val_y_t_layer_loss: 0.0597 - val_y_q_layer_root_mean_squared_error: 0.7742 - val_y_t_layer_accuracy: 0.9884\n","Epoch 36/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.6588 - y_q_layer_loss: 0.6053 - y_t_layer_loss: 0.0535 - y_q_layer_root_mean_squared_error: 0.7780 - y_t_layer_accuracy: 0.9911 - val_loss: 0.6481 - val_y_q_layer_loss: 0.5900 - val_y_t_layer_loss: 0.0581 - val_y_q_layer_root_mean_squared_error: 0.7681 - val_y_t_layer_accuracy: 0.9884\n","Epoch 37/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.6485 - y_q_layer_loss: 0.5965 - y_t_layer_loss: 0.0520 - y_q_layer_root_mean_squared_error: 0.7723 - y_t_layer_accuracy: 0.9911 - val_loss: 0.6455 - val_y_q_layer_loss: 0.5889 - val_y_t_layer_loss: 0.0566 - val_y_q_layer_root_mean_squared_error: 0.7674 - val_y_t_layer_accuracy: 0.9884\n","Epoch 38/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.6383 - y_q_layer_loss: 0.5875 - y_t_layer_loss: 0.0508 - y_q_layer_root_mean_squared_error: 0.7665 - y_t_layer_accuracy: 0.9913 - val_loss: 0.6308 - val_y_q_layer_loss: 0.5754 - val_y_t_layer_loss: 0.0553 - val_y_q_layer_root_mean_squared_error: 0.7586 - val_y_t_layer_accuracy: 0.9884\n","Epoch 39/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.6287 - y_q_layer_loss: 0.5791 - y_t_layer_loss: 0.0496 - y_q_layer_root_mean_squared_error: 0.7610 - y_t_layer_accuracy: 0.9913 - val_loss: 0.6209 - val_y_q_layer_loss: 0.5667 - val_y_t_layer_loss: 0.0542 - val_y_q_layer_root_mean_squared_error: 0.7528 - val_y_t_layer_accuracy: 0.9884\n","Epoch 40/40\n","130/130 [==============================] - 0s 3ms/step - loss: 0.6194 - y_q_layer_loss: 0.5708 - y_t_layer_loss: 0.0486 - y_q_layer_root_mean_squared_error: 0.7555 - y_t_layer_accuracy: 0.9913 - val_loss: 0.6167 - val_y_q_layer_loss: 0.5636 - val_y_t_layer_loss: 0.0531 - val_y_q_layer_root_mean_squared_error: 0.7507 - val_y_t_layer_accuracy: 0.9884\n"]}],"source":["history = model.fit(train_X, train_y, \n","                    epochs = 40, validation_data=(val_X, val_y))"]},{"cell_type":"markdown","id":"2868e18a","metadata":{"papermill":{"duration":0.04555,"end_time":"2023-07-23T13:51:32.703449","exception":false,"start_time":"2023-07-23T13:51:32.657899","status":"completed"},"tags":[]},"source":["## Metrics"]},{"cell_type":"code","execution_count":20,"id":"1e2e5d37","metadata":{"execution":{"iopub.execute_input":"2023-07-23T13:51:32.797145Z","iopub.status.busy":"2023-07-23T13:51:32.795873Z","iopub.status.idle":"2023-07-23T13:51:32.943759Z","shell.execute_reply":"2023-07-23T13:51:32.942148Z"},"papermill":{"duration":0.197247,"end_time":"2023-07-23T13:51:32.946222","exception":false,"start_time":"2023-07-23T13:51:32.748975","status":"completed"},"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["33/33 [==============================] - 0s 2ms/step - loss: 0.6167 - y_q_layer_loss: 0.5636 - y_t_layer_loss: 0.0531 - y_q_layer_root_mean_squared_error: 0.7507 - y_t_layer_accuracy: 0.9884\n","\n","loss: 0.6167219281196594\n","wine_quality_loss: 0.5635987520217896\n","wine_type_loss: 0.053123317658901215\n","wine_quality_rmse: 0.7507321238517761\n","wine_type_accuracy: 0.988394558429718\n"]}],"source":["loss, wine_quality_loss, wine_type_loss, wine_quality_rmse, wine_type_accuracy = model.evaluate(x=val_X, y=val_y)\n","\n","print()\n","print(f'loss: {loss}')\n","print(f'wine_quality_loss: {wine_quality_loss}')\n","print(f'wine_type_loss: {wine_type_loss}')\n","print(f'wine_quality_rmse: {wine_quality_rmse}')\n","print(f'wine_type_accuracy: {wine_type_accuracy}')"]},{"cell_type":"markdown","id":"17218af4","metadata":{"execution":{"iopub.execute_input":"2022-10-06T09:23:56.562669Z","iopub.status.busy":"2022-10-06T09:23:56.562219Z","iopub.status.idle":"2022-10-06T09:23:56.568081Z","shell.execute_reply":"2022-10-06T09:23:56.566793Z","shell.execute_reply.started":"2022-10-06T09:23:56.562607Z"},"papermill":{"duration":0.045046,"end_time":"2023-07-23T13:51:33.036633","exception":false,"start_time":"2023-07-23T13:51:32.991587","status":"completed"},"tags":[]},"source":["# Conclusions, Fork & Improve\n","I realy like this kind of models! They can save a lot of time and money for models in production that must be retrained daily, and that are using the same data to predict more that one variable. \n","\n","In the Model I used an additional Dense layer that maybe don't add a lot of value to the prediction but I want to show that it's possible create different branches inside the same model. \n","\n","It's an easy and powerful technique! I hope that you haver enjoyed it.\n","\n","### Feel free to fork or copy the Notebook, if you like it please upvote it. It will encourage me a lot to share more work like this one. \n","\n","There are a lot of ways to improve the result obtained with this notebook. \n","\n","Of course that you can buind a completly new model changing the layers. \n","\n","You can use callback funtions: to save the best model, to stop early the training. But be carefull because now you have more variables to watch, you need to decide with one to use in the callback functions.\n","\n","It's possible to use a different learning rate, or modify it in a callback function too. \n","\n","If you fork the model and get a different result I will be really happy to discuss it in the comments section :-) "]},{"cell_type":"markdown","id":"a8a2a8ab","metadata":{"execution":{"iopub.execute_input":"2022-10-06T09:37:38.571827Z","iopub.status.busy":"2022-10-06T09:37:38.571429Z","iopub.status.idle":"2022-10-06T09:37:38.577076Z","shell.execute_reply":"2022-10-06T09:37:38.575814Z","shell.execute_reply.started":"2022-10-06T09:37:38.571796Z"},"papermill":{"duration":0.044771,"end_time":"2023-07-23T13:51:33.126778","exception":false,"start_time":"2023-07-23T13:51:33.082007","status":"completed"},"tags":[]},"source":["# Inspirations\n","The only inspiration is a course tat I'm following on Coursera: Advanced Techniques with Tensorflow. Expect more notebooks like this one with more advanced topics as I'm advancing with the specialization. \n","\n","There are not so much models with Multiple Outputs in Kaggle, sincerely I can't found any actual. \n","\n","Just a special mention to the Dataset Used: \n","https://www.kaggle.com/datasets/rajyellow46/wine-quality\n","\n","Datased original in: \n","Dua, D. and Graff, C. (2019). UCI Machine Learning Repository [http://archive.ics.uci.edu/ml]. Irvine, CA: University of California, School of Information and Computer Science."]},{"cell_type":"markdown","id":"37469dbd","metadata":{"papermill":{"duration":0.044703,"end_time":"2023-07-23T13:51:33.218533","exception":false,"start_time":"2023-07-23T13:51:33.17383","status":"completed"},"tags":[]},"source":["# More Notebooks in the TensorFlow Serie. \n","I'm working in a serie of notebooks with some interesting techniques in Tensorflow: Tensorflow beyond the basics.\n","\n","**How to create a Siamese Network to compare images.** \n","https://www.kaggle.com/code/peremartramanonellas/how-to-create-a-siamese-network-to-compare-images\n","\n","**Improve Tensorflow performance with Graph mode.**\n","https://www.kaggle.com/code/peremartramanonellas/improve-tensorflow-performance-with-graph-mode\n","Compare the Eager vs Graph mode and improve execution of custom models and Data treatment. \n","\n","### Feel Free to fork or edit the noteboook for you own convenience. Please consider ***UPVOTING IT***. It helps others to discover the notebook, and it encourages me to continue publishing."]},{"cell_type":"code","execution_count":null,"id":"cf3ef99e","metadata":{"papermill":{"duration":0.045137,"end_time":"2023-07-23T13:51:33.308663","exception":false,"start_time":"2023-07-23T13:51:33.263526","status":"completed"},"tags":[]},"outputs":[],"source":[]}],"metadata":{"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"codemirror_mode":{"name":"ipython","version":3},"file_extension":".py","mimetype":"text/x-python","name":"python","nbconvert_exporter":"python","pygments_lexer":"ipython3","version":"3.7.12"},"papermill":{"default_parameters":{},"duration":44.847436,"end_time":"2023-07-23T13:51:36.34139","environment_variables":{},"exception":null,"input_path":"__notebook__.ipynb","output_path":"__notebook__.ipynb","parameters":{},"start_time":"2023-07-23T13:50:51.493954","version":"2.3.4"}},"nbformat":4,"nbformat_minor":5}