{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Neural Network Model Trainning"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Module Dependencies"
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"# Import\n",
"import pandas as pd\n",
"import tensorflow as tf\n",
"import keras_tuner as kt\n",
"import matplotlib.pyplot as plt\n",
"from sklearn.preprocessing import StandardScaler\n",
"from sklearn.model_selection import train_test_split\n",
"from tensorflow.keras.utils import to_categorical\n",
"\n",
"# Set the option to display all columns\n",
"pd.set_option(\"display.max_columns\", None)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Data Loading"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"
\n",
"\n",
"
\n",
" \n",
" \n",
" | \n",
" Person ID | \n",
" Gender | \n",
" Age | \n",
" Occupation | \n",
" Sleep Duration | \n",
" Quality of Sleep | \n",
" Physical Activity Level | \n",
" Stress Level | \n",
" BMI Category | \n",
" Blood Pressure | \n",
" Heart Rate | \n",
" Daily Steps | \n",
" Sleep Disorder | \n",
"
\n",
" \n",
" \n",
" \n",
" | 0 | \n",
" 1 | \n",
" Male | \n",
" 27 | \n",
" Software Engineer | \n",
" 6.1 | \n",
" 6 | \n",
" 42 | \n",
" 6 | \n",
" Overweight | \n",
" 126/83 | \n",
" 77 | \n",
" 4200 | \n",
" NaN | \n",
"
\n",
" \n",
" | 1 | \n",
" 2 | \n",
" Male | \n",
" 28 | \n",
" Doctor | \n",
" 6.2 | \n",
" 6 | \n",
" 60 | \n",
" 8 | \n",
" Normal | \n",
" 125/80 | \n",
" 75 | \n",
" 10000 | \n",
" NaN | \n",
"
\n",
" \n",
" | 2 | \n",
" 3 | \n",
" Male | \n",
" 28 | \n",
" Doctor | \n",
" 6.2 | \n",
" 6 | \n",
" 60 | \n",
" 8 | \n",
" Normal | \n",
" 125/80 | \n",
" 75 | \n",
" 10000 | \n",
" NaN | \n",
"
\n",
" \n",
" | 3 | \n",
" 4 | \n",
" Male | \n",
" 28 | \n",
" Sales Representative | \n",
" 5.9 | \n",
" 4 | \n",
" 30 | \n",
" 8 | \n",
" Obese | \n",
" 140/90 | \n",
" 85 | \n",
" 3000 | \n",
" Sleep Apnea | \n",
"
\n",
" \n",
" | 4 | \n",
" 5 | \n",
" Male | \n",
" 28 | \n",
" Sales Representative | \n",
" 5.9 | \n",
" 4 | \n",
" 30 | \n",
" 8 | \n",
" Obese | \n",
" 140/90 | \n",
" 85 | \n",
" 3000 | \n",
" Sleep Apnea | \n",
"
\n",
" \n",
"
\n",
"
"
],
"text/plain": [
" Person ID Gender Age Occupation Sleep Duration \\\n",
"0 1 Male 27 Software Engineer 6.1 \n",
"1 2 Male 28 Doctor 6.2 \n",
"2 3 Male 28 Doctor 6.2 \n",
"3 4 Male 28 Sales Representative 5.9 \n",
"4 5 Male 28 Sales Representative 5.9 \n",
"\n",
" Quality of Sleep Physical Activity Level Stress Level BMI Category \\\n",
"0 6 42 6 Overweight \n",
"1 6 60 8 Normal \n",
"2 6 60 8 Normal \n",
"3 4 30 8 Obese \n",
"4 4 30 8 Obese \n",
"\n",
" Blood Pressure Heart Rate Daily Steps Sleep Disorder \n",
"0 126/83 77 4200 NaN \n",
"1 125/80 75 10000 NaN \n",
"2 125/80 75 10000 NaN \n",
"3 140/90 85 3000 Sleep Apnea \n",
"4 140/90 85 3000 Sleep Apnea "
]
},
"execution_count": 2,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# Load the dataset\n",
"file_path = \"Resources/Sleep_health_and_lifestyle_dataset.csv\"\n",
"df = pd.read_csv(file_path)\n",
"\n",
"# Show the original DataFrame\n",
"df.head()"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n",
"RangeIndex: 374 entries, 0 to 373\n",
"Data columns (total 13 columns):\n",
" # Column Non-Null Count Dtype \n",
"--- ------ -------------- ----- \n",
" 0 Person ID 374 non-null int64 \n",
" 1 Gender 374 non-null object \n",
" 2 Age 374 non-null int64 \n",
" 3 Occupation 374 non-null object \n",
" 4 Sleep Duration 374 non-null float64\n",
" 5 Quality of Sleep 374 non-null int64 \n",
" 6 Physical Activity Level 374 non-null int64 \n",
" 7 Stress Level 374 non-null int64 \n",
" 8 BMI Category 374 non-null object \n",
" 9 Blood Pressure 374 non-null object \n",
" 10 Heart Rate 374 non-null int64 \n",
" 11 Daily Steps 374 non-null int64 \n",
" 12 Sleep Disorder 155 non-null object \n",
"dtypes: float64(1), int64(7), object(5)\n",
"memory usage: 38.1+ KB\n"
]
}
],
"source": [
"# Show the DataFrame basic information\n",
"df.info()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Data Preprocessing"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"\n",
"\n",
"
\n",
" \n",
" \n",
" | \n",
" Gender | \n",
" Age | \n",
" Occupation | \n",
" Sleep Duration | \n",
" Quality of Sleep | \n",
" Physical Activity Level | \n",
" Stress Level | \n",
" BMI Category | \n",
" Heart Rate | \n",
" Daily Steps | \n",
" Sleep Disorder | \n",
" Systolic | \n",
" Diastolic | \n",
"
\n",
" \n",
" \n",
" \n",
" | 0 | \n",
" Male | \n",
" 27 | \n",
" Software Engineer | \n",
" 6.1 | \n",
" 6 | \n",
" 42 | \n",
" 6 | \n",
" Overweight | \n",
" 77 | \n",
" 4200 | \n",
" 0.0 | \n",
" 126 | \n",
" 83 | \n",
"
\n",
" \n",
" | 1 | \n",
" Male | \n",
" 28 | \n",
" Doctor | \n",
" 6.2 | \n",
" 6 | \n",
" 60 | \n",
" 8 | \n",
" Normal | \n",
" 75 | \n",
" 10000 | \n",
" 0.0 | \n",
" 125 | \n",
" 80 | \n",
"
\n",
" \n",
" | 2 | \n",
" Male | \n",
" 28 | \n",
" Doctor | \n",
" 6.2 | \n",
" 6 | \n",
" 60 | \n",
" 8 | \n",
" Normal | \n",
" 75 | \n",
" 10000 | \n",
" 0.0 | \n",
" 125 | \n",
" 80 | \n",
"
\n",
" \n",
" | 3 | \n",
" Male | \n",
" 28 | \n",
" Sales Representative | \n",
" 5.9 | \n",
" 4 | \n",
" 30 | \n",
" 8 | \n",
" Obese | \n",
" 85 | \n",
" 3000 | \n",
" 2.0 | \n",
" 140 | \n",
" 90 | \n",
"
\n",
" \n",
" | 4 | \n",
" Male | \n",
" 28 | \n",
" Sales Representative | \n",
" 5.9 | \n",
" 4 | \n",
" 30 | \n",
" 8 | \n",
" Obese | \n",
" 85 | \n",
" 3000 | \n",
" 2.0 | \n",
" 140 | \n",
" 90 | \n",
"
\n",
" \n",
"
\n",
"
"
],
"text/plain": [
" Gender Age Occupation Sleep Duration Quality of Sleep \\\n",
"0 Male 27 Software Engineer 6.1 6 \n",
"1 Male 28 Doctor 6.2 6 \n",
"2 Male 28 Doctor 6.2 6 \n",
"3 Male 28 Sales Representative 5.9 4 \n",
"4 Male 28 Sales Representative 5.9 4 \n",
"\n",
" Physical Activity Level Stress Level BMI Category Heart Rate \\\n",
"0 42 6 Overweight 77 \n",
"1 60 8 Normal 75 \n",
"2 60 8 Normal 75 \n",
"3 30 8 Obese 85 \n",
"4 30 8 Obese 85 \n",
"\n",
" Daily Steps Sleep Disorder Systolic Diastolic \n",
"0 4200 0.0 126 83 \n",
"1 10000 0.0 125 80 \n",
"2 10000 0.0 125 80 \n",
"3 3000 2.0 140 90 \n",
"4 3000 2.0 140 90 "
]
},
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# Remove the \"Person ID\" column\n",
"df.drop(\"Person ID\", axis=1, inplace=True)\n",
"\n",
"# Split the \"Blood Pressure\" column into \"Systolic\" and \"Diastolic\"\n",
"df[\"Systolic\"] = df[\"Blood Pressure\"].apply(lambda x: int(x.split(\"/\")[0]))\n",
"df[\"Diastolic\"] = df[\"Blood Pressure\"].apply(lambda x: int(x.split(\"/\")[1]))\n",
"df.drop(\"Blood Pressure\", axis=1, inplace=True)\n",
"\n",
"# Replace the values on the \"Sleep Disorder\" column with the value 1 for \"Insombia\" and the value 2 for \"Sleep Apnea\"\n",
"df[\"Sleep Disorder\"] = df[\"Sleep Disorder\"].map({\"Insomnia\": 1, \"Sleep Apnea\": 2})\n",
"\n",
"# Fill the \"blanks\" of the \"Sleep Disorder\" column with the value 0\n",
"df[\"Sleep Disorder\"] = df[\"Sleep Disorder\"].fillna(0)\n",
"\n",
"# Show the modified DataFrame\n",
"df.head()"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [],
"source": [
"# Define the Machine Learning Model Features and Output\n",
"X = df.drop(\"Sleep Disorder\", axis=1)\n",
"y = df[\"Sleep Disorder\"]\n",
"\n",
"# Identify the categorical and numerical columns of the \"X\" into lists\n",
"categorical_columns = [\"Gender\", \"Occupation\", \"BMI Category\"]\n",
"numerical_columns = [\"Age\", \"Sleep Duration\", \"Quality of Sleep\", \"Physical Activity Level\", \"Stress Level\", \"Heart Rate\", \"Daily Steps\", \"Systolic\", \"Diastolic\"]\n",
"\n",
"# Transform the categorical columns of the DataFrame into dummy/indicator variables using the function get_dummies()\n",
"X = pd.get_dummies(X, columns=categorical_columns, drop_first=True)\n",
"\n",
"# Split the datasets into trainning and test data\n",
"X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)\n",
"\n",
"# Define the scaling parameters of numerical columns using the trainning data set\n",
"scaler = StandardScaler().fit(X_train[numerical_columns])"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"\n",
"\n",
"
\n",
" \n",
" \n",
" | \n",
" Age | \n",
" Sleep Duration | \n",
" Quality of Sleep | \n",
" Physical Activity Level | \n",
" Stress Level | \n",
" Heart Rate | \n",
" Daily Steps | \n",
" Systolic | \n",
" Diastolic | \n",
" Gender_Male | \n",
" Occupation_Doctor | \n",
" Occupation_Engineer | \n",
" Occupation_Lawyer | \n",
" Occupation_Manager | \n",
" Occupation_Nurse | \n",
" Occupation_Sales Representative | \n",
" Occupation_Salesperson | \n",
" Occupation_Scientist | \n",
" Occupation_Software Engineer | \n",
" Occupation_Teacher | \n",
" BMI Category_Normal Weight | \n",
" BMI Category_Obese | \n",
" BMI Category_Overweight | \n",
"
\n",
" \n",
" \n",
" \n",
" | 192 | \n",
" 0.055811 | \n",
" -0.833457 | \n",
" -1.155098 | \n",
" -0.728644 | \n",
" 0.956601 | \n",
" 0.481081 | \n",
" -0.554643 | \n",
" 0.186918 | \n",
" 0.050678 | \n",
" True | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" True | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" True | \n",
"
\n",
" \n",
" | 75 | \n",
" -1.111141 | \n",
" -1.468533 | \n",
" -1.155098 | \n",
" -1.450369 | \n",
" 1.522984 | \n",
" 0.481081 | \n",
" -1.180447 | \n",
" -0.448177 | \n",
" -0.746830 | \n",
" True | \n",
" True | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
"
\n",
" \n",
" | 84 | \n",
" -0.877751 | \n",
" 0.436694 | \n",
" 0.537913 | \n",
" -0.006920 | \n",
" -0.176166 | \n",
" -0.014076 | \n",
" 0.696966 | \n",
" -1.083272 | \n",
" -0.746830 | \n",
" True | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" True | \n",
" False | \n",
" True | \n",
" False | \n",
" False | \n",
"
\n",
" \n",
" | 362 | \n",
" 1.922934 | \n",
" 1.325801 | \n",
" 1.384419 | \n",
" 0.714805 | \n",
" -1.308933 | \n",
" -0.509234 | \n",
" 0.071162 | \n",
" 1.457107 | \n",
" 1.645692 | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" True | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" True | \n",
"
\n",
" \n",
" | 16 | \n",
" -1.577922 | \n",
" -0.833457 | \n",
" -2.001604 | \n",
" -0.969219 | \n",
" 0.956601 | \n",
" 2.461710 | \n",
" -1.806251 | \n",
" 0.440955 | \n",
" 0.369681 | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" True | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" False | \n",
" True | \n",
" False | \n",
" False | \n",
"
\n",
" \n",
"
\n",
"
"
],
"text/plain": [
" Age Sleep Duration Quality of Sleep Physical Activity Level \\\n",
"192 0.055811 -0.833457 -1.155098 -0.728644 \n",
"75 -1.111141 -1.468533 -1.155098 -1.450369 \n",
"84 -0.877751 0.436694 0.537913 -0.006920 \n",
"362 1.922934 1.325801 1.384419 0.714805 \n",
"16 -1.577922 -0.833457 -2.001604 -0.969219 \n",
"\n",
" Stress Level Heart Rate Daily Steps Systolic Diastolic Gender_Male \\\n",
"192 0.956601 0.481081 -0.554643 0.186918 0.050678 True \n",
"75 1.522984 0.481081 -1.180447 -0.448177 -0.746830 True \n",
"84 -0.176166 -0.014076 0.696966 -1.083272 -0.746830 True \n",
"362 -1.308933 -0.509234 0.071162 1.457107 1.645692 False \n",
"16 0.956601 2.461710 -1.806251 0.440955 0.369681 False \n",
"\n",
" Occupation_Doctor Occupation_Engineer Occupation_Lawyer \\\n",
"192 False False False \n",
"75 True False False \n",
"84 False False False \n",
"362 False False False \n",
"16 False False False \n",
"\n",
" Occupation_Manager Occupation_Nurse Occupation_Sales Representative \\\n",
"192 False False False \n",
"75 False False False \n",
"84 False False False \n",
"362 False True False \n",
"16 False True False \n",
"\n",
" Occupation_Salesperson Occupation_Scientist \\\n",
"192 True False \n",
"75 False False \n",
"84 False False \n",
"362 False False \n",
"16 False False \n",
"\n",
" Occupation_Software Engineer Occupation_Teacher \\\n",
"192 False False \n",
"75 False False \n",
"84 True False \n",
"362 False False \n",
"16 False False \n",
"\n",
" BMI Category_Normal Weight BMI Category_Obese BMI Category_Overweight \n",
"192 False False True \n",
"75 False False False \n",
"84 True False False \n",
"362 False False True \n",
"16 True False False "
]
},
"execution_count": 6,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# Scale the values of the \"X\" sets\n",
"X_train[numerical_columns] = scaler.transform(X_train[numerical_columns])\n",
"X_test[numerical_columns] = scaler.transform(X_test[numerical_columns])\n",
"\n",
"# Show the trainning set\n",
"X_train.head()"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"[[0. 1. 0.]\n",
" [1. 0. 0.]\n",
" [1. 0. 0.]\n",
" [0. 0. 1.]\n",
" [0. 0. 1.]]\n"
]
}
],
"source": [
"# Use the function to_categorical to encode the integer labels of the \"y\" sets\n",
"y_train = to_categorical(y_train, 3)\n",
"y_test = to_categorical(y_test, 3)\n",
"\n",
"# Show \"y_train\"\n",
"print(y_train[:5])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Keras Tuner Setting"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {},
"outputs": [],
"source": [
"# Define the number of input parameters\n",
"number_input_features = len(X_train.columns)\n",
"\n",
"# Create a method for the hyperparameters optimization of a Sequential Model\n",
"\n",
"def model_optimization(hp):\n",
"\n",
" # Initialize the sequential model\n",
" nn_model = tf.keras.models.Sequential()\n",
"\n",
" # Define the activation function options to use in the hidden layers\n",
" activation_function_choice = hp.Choice('activation',['relu','tanh','sigmoid'])\n",
"\n",
" # Add the first hidden layer of the model\n",
" nn_model.add(tf.keras.layers.Dense(\n",
" units = hp.Int(\"first_units\", min_value = 1, max_value = 10, step = 2), # Allow keras tuner to decide the number of neurons on the first layer\n",
" input_dim = number_input_features,\n",
" activation = activation_function_choice))\n",
" \n",
" # Allow keras tuner to decide the number of hidden layers between\n",
" for i in range(hp.Int(\"num_layers\", 1, 10)):\n",
" nn_model.add(tf.keras.layers.Dense(\n",
" units = hp.Int(\"units_\" + str(i), min_value = 1, max_value = 10, step = 2), # Allow keras tuner to decide the number of neurons on the hidden layers\n",
" activation = activation_function_choice))\n",
"\n",
" # Add the output layer for the 3 different labels of \"Sleep Disorder\"\n",
" nn_model.add(tf.keras.layers.Dense(units=3, activation=\"softmax\"))\n",
"\n",
" # Compile the Sequential model together and customize metrics\n",
" nn_model.compile(loss=\"categorical_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\"])\n",
"\n",
" return nn_model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Hyperparameter Evaluation"
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Trial 178 Complete [00h 00m 08s]\n",
"val_accuracy: 0.5733333230018616\n",
"\n",
"Best val_accuracy So Far: 0.9333333373069763\n",
"Total elapsed time: 00h 15m 37s\n"
]
}
],
"source": [
"# Define the keras tuner\n",
"tuner = kt.Hyperband(\n",
" model_optimization,\n",
" objective=\"val_accuracy\",\n",
" max_epochs=50,\n",
" hyperband_iterations=2)\n",
"\n",
"# Run the kerastuner search for best hyperparameters\n",
"tuner.search(X_train, y_train, epochs=40, validation_data = (X_test, y_test))"
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"{'activation': 'tanh',\n",
" 'first_units': 9,\n",
" 'num_layers': 1,\n",
" 'units_0': 7,\n",
" 'units_1': 3,\n",
" 'units_2': 5,\n",
" 'units_3': 1,\n",
" 'units_4': 3,\n",
" 'units_5': 7,\n",
" 'units_6': 9,\n",
" 'units_7': 1,\n",
" 'units_8': 9,\n",
" 'units_9': 3,\n",
" 'tuner/epochs': 50,\n",
" 'tuner/initial_epoch': 0,\n",
" 'tuner/bracket': 0,\n",
" 'tuner/round': 0}"
]
},
"execution_count": 10,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# Get best model hyperparameters\n",
"best_hyper = tuner.get_best_hyperparameters(1)[0]\n",
"best_hyper.values"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Model with Optimized Hyperparameters"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Model Setup"
]
},
{
"cell_type": "code",
"execution_count": 11,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"c:\\Users\\bradl\\anaconda3\\envs\\dev3.10\\lib\\site-packages\\keras\\src\\layers\\core\\dense.py:87: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.\n",
" super().__init__(activity_regularizer=activity_regularizer, **kwargs)\n"
]
},
{
"data": {
"text/html": [
"Model: \"sequential_1\"\n",
"\n"
],
"text/plain": [
"\u001b[1mModel: \"sequential_1\"\u001b[0m\n"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
"┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓\n",
"┃ Layer (type) ┃ Output Shape ┃ Param # ┃\n",
"┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩\n",
"│ dense_12 (Dense) │ (None, 5) │ 120 │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ dense_13 (Dense) │ (None, 5) │ 30 │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ dense_14 (Dense) │ (None, 7) │ 42 │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ dense_15 (Dense) │ (None, 3) │ 24 │\n",
"└─────────────────────────────────┴────────────────────────┴───────────────┘\n",
"\n"
],
"text/plain": [
"┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓\n",
"┃\u001b[1m \u001b[0m\u001b[1mLayer (type) \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1mOutput Shape \u001b[0m\u001b[1m \u001b[0m┃\u001b[1m \u001b[0m\u001b[1m Param #\u001b[0m\u001b[1m \u001b[0m┃\n",
"┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩\n",
"│ dense_12 (\u001b[38;5;33mDense\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m5\u001b[0m) │ \u001b[38;5;34m120\u001b[0m │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ dense_13 (\u001b[38;5;33mDense\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m5\u001b[0m) │ \u001b[38;5;34m30\u001b[0m │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ dense_14 (\u001b[38;5;33mDense\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m7\u001b[0m) │ \u001b[38;5;34m42\u001b[0m │\n",
"├─────────────────────────────────┼────────────────────────┼───────────────┤\n",
"│ dense_15 (\u001b[38;5;33mDense\u001b[0m) │ (\u001b[38;5;45mNone\u001b[0m, \u001b[38;5;34m3\u001b[0m) │ \u001b[38;5;34m24\u001b[0m │\n",
"└─────────────────────────────────┴────────────────────────┴───────────────┘\n"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" Total params: 216 (864.00 B)\n",
"\n"
],
"text/plain": [
"\u001b[1m Total params: \u001b[0m\u001b[38;5;34m216\u001b[0m (864.00 B)\n"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" Trainable params: 216 (864.00 B)\n",
"\n"
],
"text/plain": [
"\u001b[1m Trainable params: \u001b[0m\u001b[38;5;34m216\u001b[0m (864.00 B)\n"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"text/html": [
" Non-trainable params: 0 (0.00 B)\n",
"\n"
],
"text/plain": [
"\u001b[1m Non-trainable params: \u001b[0m\u001b[38;5;34m0\u001b[0m (0.00 B)\n"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# Initialize the sequential neural network model\n",
"nn_model = tf.keras.models.Sequential()\n",
"\n",
"# Add the first hidden layer of the model\n",
"nn_model.add(tf.keras.layers.Dense(units=5, input_dim=number_input_features, activation=\"tanh\"))\n",
"\n",
"# Add the second hidden layer of the model\n",
"nn_model.add(tf.keras.layers.Dense(units=5, activation=\"tanh\"))\n",
"\n",
"# Add the third hidden layer of the model\n",
"nn_model.add(tf.keras.layers.Dense(units=7, activation=\"tanh\"))\n",
"\n",
"# Add the output layer for the 3 different labels of \"Sleep Disorder\"\n",
"nn_model.add(tf.keras.layers.Dense(units=3, activation=\"softmax\"))\n",
"\n",
"# Check the structure of the model\n",
"nn_model.summary()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Compile and Train the Model"
]
},
{
"cell_type": "code",
"execution_count": 12,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 2ms/step - accuracy: 0.2074 - loss: 1.1753 \n",
"Epoch 2/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.3117 - loss: 1.1062 \n",
"Epoch 3/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.4320 - loss: 1.0627 \n",
"Epoch 4/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.4936 - loss: 0.9977 \n",
"Epoch 5/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.6234 - loss: 0.9472 \n",
"Epoch 6/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.7389 - loss: 0.8854 \n",
"Epoch 7/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.7345 - loss: 0.8763 \n",
"Epoch 8/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.7437 - loss: 0.8313 \n",
"Epoch 9/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.7395 - loss: 0.7972 \n",
"Epoch 10/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.7313 - loss: 0.7763 \n",
"Epoch 11/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.7470 - loss: 0.7556 \n",
"Epoch 12/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.7252 - loss: 0.7412 \n",
"Epoch 13/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.7638 - loss: 0.6890 \n",
"Epoch 14/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.7557 - loss: 0.6685 \n",
"Epoch 15/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.7732 - loss: 0.6175 \n",
"Epoch 16/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.7933 - loss: 0.6282 \n",
"Epoch 17/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.8501 - loss: 0.5804 \n",
"Epoch 18/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.8052 - loss: 0.6257 \n",
"Epoch 19/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.8160 - loss: 0.5945 \n",
"Epoch 20/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.8638 - loss: 0.5464 \n",
"Epoch 21/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.8845 - loss: 0.5335 \n",
"Epoch 22/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.8814 - loss: 0.5702 \n",
"Epoch 23/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.8855 - loss: 0.5220 \n",
"Epoch 24/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.8873 - loss: 0.5152 \n",
"Epoch 25/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.8983 - loss: 0.5115 \n",
"Epoch 26/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.8752 - loss: 0.5194 \n",
"Epoch 27/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.8842 - loss: 0.4921 \n",
"Epoch 28/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9037 - loss: 0.4886 \n",
"Epoch 29/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.8883 - loss: 0.5263 \n",
"Epoch 30/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9139 - loss: 0.4587 \n",
"Epoch 31/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9280 - loss: 0.4008 \n",
"Epoch 32/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.8812 - loss: 0.5104 \n",
"Epoch 33/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9000 - loss: 0.4473 \n",
"Epoch 34/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9027 - loss: 0.4366 \n",
"Epoch 35/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9140 - loss: 0.4082 \n",
"Epoch 36/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.8940 - loss: 0.4539 \n",
"Epoch 37/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 6ms/step - accuracy: 0.8923 - loss: 0.4404\n",
"Epoch 38/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.8884 - loss: 0.4600 \n",
"Epoch 39/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.8944 - loss: 0.4290 \n",
"Epoch 40/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9249 - loss: 0.3834 \n",
"Epoch 41/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.8997 - loss: 0.3998 \n",
"Epoch 42/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9207 - loss: 0.3639 \n",
"Epoch 43/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.8908 - loss: 0.4325 \n",
"Epoch 44/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.8924 - loss: 0.4317 \n",
"Epoch 45/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9039 - loss: 0.3706 \n",
"Epoch 46/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9085 - loss: 0.3748 \n",
"Epoch 47/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9125 - loss: 0.3595 \n",
"Epoch 48/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9093 - loss: 0.3714 \n",
"Epoch 49/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9071 - loss: 0.3518 \n",
"Epoch 50/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9104 - loss: 0.3675 \n",
"Epoch 51/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9195 - loss: 0.3338 \n",
"Epoch 52/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9286 - loss: 0.3233 \n",
"Epoch 53/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9010 - loss: 0.3736 \n",
"Epoch 54/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9098 - loss: 0.3702 \n",
"Epoch 55/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9094 - loss: 0.3739 \n",
"Epoch 56/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9182 - loss: 0.3382 \n",
"Epoch 57/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9313 - loss: 0.3043 \n",
"Epoch 58/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9145 - loss: 0.3400 \n",
"Epoch 59/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9040 - loss: 0.3525 \n",
"Epoch 60/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9098 - loss: 0.3642 \n",
"Epoch 61/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9213 - loss: 0.3083 \n",
"Epoch 62/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9173 - loss: 0.3619 \n",
"Epoch 63/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9157 - loss: 0.3475 \n",
"Epoch 64/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9013 - loss: 0.3918 \n",
"Epoch 65/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9100 - loss: 0.3294 \n",
"Epoch 66/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9174 - loss: 0.3587 \n",
"Epoch 67/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9185 - loss: 0.3278 \n",
"Epoch 68/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9392 - loss: 0.2909 \n",
"Epoch 69/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9231 - loss: 0.3332 \n",
"Epoch 70/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9071 - loss: 0.3403 \n",
"Epoch 71/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9427 - loss: 0.2723 \n",
"Epoch 72/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9296 - loss: 0.2920 \n",
"Epoch 73/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9379 - loss: 0.2867 \n",
"Epoch 74/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9169 - loss: 0.3458 \n",
"Epoch 75/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9305 - loss: 0.2846 \n",
"Epoch 76/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9232 - loss: 0.3425 \n",
"Epoch 77/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9203 - loss: 0.2927 \n",
"Epoch 78/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9413 - loss: 0.2680 \n",
"Epoch 79/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9279 - loss: 0.3206 \n",
"Epoch 80/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9164 - loss: 0.3357 \n",
"Epoch 81/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9129 - loss: 0.3682 \n",
"Epoch 82/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9151 - loss: 0.3164 \n",
"Epoch 83/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9280 - loss: 0.3237 \n",
"Epoch 84/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9058 - loss: 0.4000 \n",
"Epoch 85/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9214 - loss: 0.3349 \n",
"Epoch 86/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9205 - loss: 0.3352 \n",
"Epoch 87/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9235 - loss: 0.3211 \n",
"Epoch 88/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9259 - loss: 0.2963 \n",
"Epoch 89/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9145 - loss: 0.3427 \n",
"Epoch 90/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9210 - loss: 0.3304 \n",
"Epoch 91/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9291 - loss: 0.3175 \n",
"Epoch 92/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9248 - loss: 0.3083 \n",
"Epoch 93/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9270 - loss: 0.3038 \n",
"Epoch 94/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9308 - loss: 0.2892 \n",
"Epoch 95/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9196 - loss: 0.3128 \n",
"Epoch 96/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9211 - loss: 0.3049 \n",
"Epoch 97/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9230 - loss: 0.3154 \n",
"Epoch 98/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9343 - loss: 0.2580 \n",
"Epoch 99/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 1ms/step - accuracy: 0.9383 - loss: 0.2586 \n",
"Epoch 100/100\n",
"\u001b[1m10/10\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m0s\u001b[0m 2ms/step - accuracy: 0.9310 - loss: 0.2803 \n"
]
}
],
"source": [
"# Compile the Sequential model together and customize metrics\n",
"nn_model.compile(loss=\"categorical_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\"])\n",
"\n",
"# Fit the model to the training data\n",
"fit_model = nn_model.fit(X_train, y_train, epochs=100)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Trainned Model Performance"
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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",
"text/plain": [
""
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# Create a DataFrame containing training history\n",
"history_df = pd.DataFrame(fit_model.history)\n",
"\n",
"# Increase the index by 1 to match the number of epochs\n",
"history_df.index += 1\n",
"\n",
"# Plot the loss and accuracy\n",
"\n",
"# Create a figure and axis\n",
"fig, ax1 = plt.subplots()\n",
"\n",
"# Plot 'loss' on the primary y-axis\n",
"ax1.plot(history_df.index, history_df[\"loss\"], color=\"blue\", label=\"Loss\")\n",
"ax1.set_xlabel(\"Epoch\")\n",
"ax1.set_ylabel(\"Loss\", color=\"blue\")\n",
"ax1.tick_params(axis=\"y\", labelcolor=\"blue\")\n",
"\n",
"# Create a secondary y-axis for 'accuracy'\n",
"ax2 = ax1.twinx()\n",
"ax2.plot(history_df.index, history_df[\"accuracy\"], color=\"green\", label=\"Accuracy\")\n",
"ax2.set_ylabel(\"Accuracy\", color=\"green\")\n",
"ax2.tick_params(axis=\"y\", labelcolor=\"green\")\n",
"\n",
"# Add a title and show the plot\n",
"plt.title(\"Loss and Accuracy Over Epochs\")\n",
"fig.tight_layout()\n",
"plt.show()"
]
},
{
"cell_type": "code",
"execution_count": 14,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"3/3 - 0s - 42ms/step - accuracy: 0.8933 - loss: 0.3771\n"
]
}
],
"source": [
"# Evaluate the model using the test data\n",
"model_loss, model_accuracy = nn_model.evaluate(X_test,y_test,verbose=2)"
]
},
{
"cell_type": "code",
"execution_count": 15,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"WARNING:absl:You are saving your model as an HDF5 file via `model.save()` or `keras.saving.save_model(model)`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')` or `keras.saving.save_model(model, 'my_model.keras')`. \n"
]
}
],
"source": [
"#Save the trained model\n",
"nn_model.save(\"sleep_disorder_model.h5\")"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "dev3.10",
"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": "3.10.14"
}
},
"nbformat": 4,
"nbformat_minor": 2
}