From 1fe4667e952ea43e39e1371591613323fcd41ccc Mon Sep 17 00:00:00 2001
From: Andreas Leitherer <leitherer@fhi-berlin.mpg.de>
Date: Mon, 4 Jan 2021 09:02:12 +0100
Subject: [PATCH] Small changes in text
---
nn_regression.ipynb | 433 ++++++++++++++++++++++++++++++++++++++++----
1 file changed, 398 insertions(+), 35 deletions(-)
diff --git a/nn_regression.ipynb b/nn_regression.ipynb
index 52d91f8..f0c4153 100644
--- a/nn_regression.ipynb
+++ b/nn_regression.ipynb
@@ -72,7 +72,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 1,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:48:01.350451Z",
@@ -80,7 +80,15 @@
},
"scrolled": true
},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stderr",
+ "output_type": "stream",
+ "text": [
+ "Using TensorFlow backend.\n"
+ ]
+ }
+ ],
"source": [
"# Plotting\n",
"%matplotlib inline\n",
@@ -161,7 +169,7 @@
"\n",
"$(\\mathbf{x}^{(1)}, y^{(1)}), (\\mathbf{x}^{(2)}, y^{(2)}), ..., (\\mathbf{x}^{(m)}, y^{(m)})$.\n",
"\n",
- "Weights and bias term of the model (here: $w_1, w_2, b$) are optimized by minimizing a *loss function* $L(w_1, w_2, b)$ that quantifies how the predicted values $\\hat{y}^{(i)}$ deviate from the true values $y^{(i)}$ - as a function of the model parameters. We will see an explicit form for $L(w_1, w_2, b)$ in context of regression in section 2 of this tutorial. Usually gradient decent is used to find the parameters minimizing $L$ (with modifications of gradient decent enabling faster convergence, see, for instance chapter 5 and 8 of [this](https://www.deeplearningbook.org/) standard reference). After finishing optimization, in case of classification, model performance can be assessed via the classification accuracy (# of correct predictions divided by # of total predictions).\n",
+ "Weights and bias term of the model (here: $w_1, w_2, b$) are optimized by minimizing a *loss function* $L(w_1, w_2, b)$ that quantifies how the predicted values $\\hat{y}^{(i)}$ deviate from the true values $y^{(i)}$ - as a function of the model parameters. We will see an explicit form for $L(w_1, w_2, b)$ in context of regression in section 2 of this tutorial. Usually gradient decent is used to find the parameters minimizing $L$ (with modifications of gradient decent enabling for instance faster convergence - see chapter 5 and 8 of [this](https://www.deeplearningbook.org/) standard reference). After finishing optimization, in case of classification, model performance can be assessed via the classification accuracy (# of correct predictions divided by # of total predictions).\n",
"\n",
"We will explain the optimization procedure in more detail in section 2 of the tutorial. In this example, one can think of the training / optimization phase as changing the model parameters such that the optimal position of a straight line (see above figure, right) is found, which serves as a decision boundary between the two classes. "
]
@@ -178,7 +186,7 @@
"\n",
"Rectified linear unit (ReLU): $f(x) = max(0, x)$\n",
"\n",
- "The ReLU activation function is most frequently used. Non-linear functions are essential to increase the space of possible (complex) functions that the model can learn. If no activation function would be used, i.e., the identity - also called *linear activation function*- the class of possible functions that the model can represent would be drastically reduced.\n",
+ "The ReLU activation function is most frequently used. Note that the use of non-linear functions is essential: if no activation function would be used, i.e., the identity - also called *linear activation function*- the class of possible functions that the model can represent would be drastically reduced.\n",
"\n",
""
]
@@ -189,7 +197,7 @@
"source": [
"### 1.2 Multilayer peceptron\n",
"\n",
- "Extending the idea of simple perceptrons, one can construct multilayer perceptrons as sequences of layers. \n",
+ "Extending the idea of simple perceptrons, one can construct multilayer perceptrons as a sequence of layers. \n",
"Each layer consists of a predefined number of neurons, where the neurons of the first layer (the *input layer*) correspond to the input features $\\mathbf{x} = (x_1, x_2, ...)$. The subsequent layers are called *hidden layers*. The individual neurons in each hidden layer are a linear combination of neurons from the previous layer. For instance, the *activation value* $a_1$ highlighted in the figure below is computed the following way:\n",
"\n",
"$\\begin{equation*}\n",
@@ -223,7 +231,19 @@
"\n",
"The final activation function $f^\\prime$ is chosen in a specific way, usually depending on the task being either regression or classification - we will come back to this later.\n",
"\n",
- "To simplify the above expression for $\\mathbf{o}$, one can change the definition of input vector and weight matrices such that the bias terms can be omitted. We denote the input vector as before and introduce weight matrices W, W$^\\prime$, which yields a more compact expression for the output: \n",
+ "To simplify the above expression for $\\mathbf{o}$, it is common to change the definition of input vector and weight matrices such that the bias terms can be omitted. To illustrate this, we consider the simplified case of two input features and two activations $a_1 = w_{11}x_1 + w_{12}x_2 + b_1$ and $a_2 = w_{21}x_1 + w_{22}x_2 + b_2$. Then, A and b are defined as \n",
+ "$\\begin{equation*}\n",
+ "A = \\begin{bmatrix}w_{11} & w_{12}\\\\w_{21} & w_{22}\\end{bmatrix}, b = \\begin{bmatrix}b_1 \\\\ b_2 \\end{bmatrix}.\n",
+ "\\end{equation*}$\n",
+ "\n",
+ "Introducing the new definitions \n",
+ "\n",
+ "$\\begin{equation*}\n",
+ "W = \\begin{bmatrix} b_1 & w_{11} & w_{12}\\\\ b_2 & w_{21} & w_{22}\\end{bmatrix}, x = \\begin{bmatrix} 1 \\\\ x_1 \\\\ x_2 \\end{bmatrix}\n",
+ "\\end{equation*}$\n",
+ "\n",
+ "allows to omit the bias term and replace $A\\mathbf{x}+\\mathbf{b}$ with $W\\mathbf{x}$. Coming back to the previous example, \n",
+ "we introduce weight matrices W, W$^\\prime$, which yields a more compact expression for the output: \n",
"\n",
"$\\begin{equation*}\n",
"\\mathbf{o} = f^\\prime (W^\\prime \\mathbf{a}) = f^\\prime (W^\\prime f(W \\mathbf{x})).\n",
@@ -259,7 +279,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
- "To illustrate the usefulness of softmax activation functions, let us consider the case of crystal-structure classification. The task is to assign the correct (symmetry) label to a given, unknown crystal structure as defined by atomic positions and chemical species. For instance, possible assignments could be face-centered-cubic, body-centered-cubic, diamond or hexagonal closed packed - a collection of structures that covers more than 80% of the elemental solids. More thorough explanations on deep learning applied to crystal-structure classification can be found [here](https://www.nature.com/articles/s41467-018-05169-6). When applying the multilayer perceptron architecture which we introduced above, each of the four output neurons correspond to a specific crystal structure. The use of the softmax activation function guarantees that all output activations sum to one, which is why the output vector $\\mathbf{o}$ can be considered as a vector of classification probabilities. For instance, if $\\mathbf{o} = (1, 0, 0, 0)$, the input structure is predicted to have fcc symmetry with 100\\% probability (see figure below). This is also called \"one-hot-encoding\" and corresponds to representing a given number N of classes in the standard basis in $\\mathbb{R}^\\text{N}$, i.e., by N vectors $e_i = (0, ...0, 1, 0, ..., 0)$, for $i=1, ..., N$ and all components of $e_i$ being zero except for the $i$th entry. \n",
+ "To illustrate the usefulness of softmax activation functions, let us consider the case of crystal-structure classification. The task is to assign the correct (symmetry) label to a given, unknown crystal structure as defined by atomic positions and chemical species. For instance, possible assignments could be face-centered-cubic, body-centered-cubic, diamond or hexagonal closed packed - a collection of structures that covers more than 80% of the elemental solids. More thorough explanations on deep learning applied to crystal-structure classification can be found [here](https://www.nature.com/articles/s41467-018-05169-6). When applying the multilayer perceptron architecture that we introduced above, each of the four output neurons correspond to a specific crystal structure. The use of the softmax activation function guarantees that all output activations sum to one, which is why the output vector $\\mathbf{o}$ can be considered as a vector of classification probabilities. For instance, if $\\mathbf{o} = (1, 0, 0, 0)$, the input structure is predicted to have fcc symmetry with 100\\% probability (see figure below). This is also called \"one-hot-encoding\" and corresponds to representing a given number of classes N in the standard basis in $\\mathbb{R}^\\text{N}$, i.e., by N vectors $e_i = (0, ...0, 1, 0, ..., 0)$, for $i=1, ..., N$ and all components of $e_i$ being zero except for the $i$th entry. \n",
"\n",
"<img src=\"./assets/nn_regression/cs_classification_first_example.png\" width=\"1700\">\n",
"\n"
@@ -303,7 +323,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
- "The deep-learning model behind \"ElemNet\" is essentially a multilayer perceptron with the input vector (the representation, i.e, the descriptor of inorganic compounds) being chosen in a very specific way: \n",
+ "The deep-learning model behind \"ElemNet\" is essentially a multilayer perceptron with the input vector being chosen in a very specific way: \n",
"Each compound is represented by a feature vector $\\mathbf{f}$ of fixed length, whose components correspond to the elements of the periodic table. They are sorted according to the atomic number Z in ascending order (i.e., the first component of $\\mathbf{f}$ corresponds to hydrogen, the second to Helium etc.).\n",
"For instance, given a binary compound $\\text{A}_x \\text{B}_y$ with $x+y=1$, all entries of $\\mathbf{f}$ are zero except those corresponding to element A and B. For these entries, the relative stoichiometric attributes x and y are assigned. In case of NaCl (rock salt), the representation would be \n",
"\n",
@@ -341,7 +361,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 2,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:08:25.380307Z",
@@ -349,7 +369,88 @@
},
"scrolled": true
},
- "outputs": [],
+ "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>vol_per_atom</th>\n",
+ " <th>composition</th>\n",
+ " <th>number_of_elements</th>\n",
+ " <th>stoichiometry_dicts</th>\n",
+ " </tr>\n",
+ " </thead>\n",
+ " <tbody>\n",
+ " <tr>\n",
+ " <th>0</th>\n",
+ " <td>17.8351</td>\n",
+ " <td>Li1</td>\n",
+ " <td>1</td>\n",
+ " <td>{'Li': 1}</td>\n",
+ " </tr>\n",
+ " <tr>\n",
+ " <th>1</th>\n",
+ " <td>22.9639</td>\n",
+ " <td>Mg1</td>\n",
+ " <td>1</td>\n",
+ " <td>{'Mg': 1}</td>\n",
+ " </tr>\n",
+ " <tr>\n",
+ " <th>2</th>\n",
+ " <td>41.4146</td>\n",
+ " <td>Kr1</td>\n",
+ " <td>1</td>\n",
+ " <td>{'Kr': 1}</td>\n",
+ " </tr>\n",
+ " <tr>\n",
+ " <th>3</th>\n",
+ " <td>32.9826</td>\n",
+ " <td>Na1</td>\n",
+ " <td>1</td>\n",
+ " <td>{'Na': 1}</td>\n",
+ " </tr>\n",
+ " <tr>\n",
+ " <th>4</th>\n",
+ " <td>15.2088</td>\n",
+ " <td>Pd1</td>\n",
+ " <td>1</td>\n",
+ " <td>{'Pd': 1}</td>\n",
+ " </tr>\n",
+ " </tbody>\n",
+ "</table>\n",
+ "</div>"
+ ],
+ "text/plain": [
+ " vol_per_atom composition number_of_elements stoichiometry_dicts\n",
+ "0 17.8351 Li1 1 {'Li': 1}\n",
+ "1 22.9639 Mg1 1 {'Mg': 1}\n",
+ "2 41.4146 Kr1 1 {'Kr': 1}\n",
+ "3 32.9826 Na1 1 {'Na': 1}\n",
+ "4 15.2088 Pd1 1 {'Pd': 1}"
+ ]
+ },
+ "execution_count": 2,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
"source": [
"df = pd.read_pickle('./data/nn_regression/OQMD_Ward_et_al_2016_df.pkl')\n",
"\n",
@@ -376,14 +477,45 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 3,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:08:25.710996Z",
"start_time": "2020-05-22T14:08:25.382690Z"
}
},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "<Figure size 432x288 with 2 Axes>"
+ ]
+ },
+ "metadata": {
+ "needs_background": "light"
+ },
+ "output_type": "display_data"
+ },
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "\n",
+ "Statistics of the target property:\n",
+ "\n",
+ "count 347227.000000\n",
+ "mean 22.157036\n",
+ "std 8.613623\n",
+ "min 2.723960\n",
+ "25% 16.116350\n",
+ "50% 20.881500\n",
+ "75% 26.504150\n",
+ "max 99.884500\n",
+ "Name: vol_per_atom, dtype: float64\n"
+ ]
+ }
+ ],
"source": [
"df.hist()\n",
"plt.show()\n",
@@ -400,14 +532,51 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 4,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:08:26.547127Z",
"start_time": "2020-05-22T14:08:25.713822Z"
}
},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Total number of datapoints: 347227\n",
+ "\n",
+ "Compounds with 1 element(s) appear 2472 times in the dataset\n",
+ "Compounds with 2 element(s) appear 99411 times in the dataset\n",
+ "Compounds with 3 element(s) appear 235658 times in the dataset\n",
+ "Compounds with 4 element(s) appear 8248 times in the dataset\n",
+ "Compounds with 5 element(s) appear 1320 times in the dataset\n",
+ "Compounds with 6 element(s) appear 112 times in the dataset\n",
+ "Compounds with 7 element(s) appear 6 times in the dataset\n",
+ "\n",
+ "The following elements (in total 89) appear in the dataset:\n",
+ "\n",
+ " ['H' 'He' 'Li' 'Be' 'B' 'C' 'N' 'O' 'F' 'Ne' 'Na' 'Mg' 'Al' 'Si' 'P' 'S'\n",
+ " 'Cl' 'Ar' 'K' 'Ca' 'Sc' 'Ti' 'V' 'Cr' 'Mn' 'Fe' 'Co' 'Ni' 'Cu' 'Zn' 'Ga'\n",
+ " 'Ge' 'As' 'Se' 'Br' 'Kr' 'Rb' 'Sr' 'Y' 'Zr' 'Nb' 'Mo' 'Tc' 'Ru' 'Rh' 'Pd'\n",
+ " 'Ag' 'Cd' 'In' 'Sn' 'Sb' 'Te' 'I' 'Xe' 'Cs' 'Ba' 'La' 'Ce' 'Pr' 'Nd' 'Pm'\n",
+ " 'Sm' 'Eu' 'Gd' 'Tb' 'Dy' 'Ho' 'Er' 'Tm' 'Yb' 'Lu' 'Hf' 'Ta' 'W' 'Re' 'Os'\n",
+ " 'Ir' 'Pt' 'Au' 'Hg' 'Tl' 'Pb' 'Bi' 'Ac' 'Th' 'Pa' 'U' 'Np' 'Pu']\n"
+ ]
+ },
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "<Figure size 1800x720 with 1 Axes>"
+ ]
+ },
+ "metadata": {
+ "needs_background": "light"
+ },
+ "output_type": "display_data"
+ }
+ ],
"source": [
"print(\"Total number of datapoints: {}\\n\".format(len(y_vol_per_atom)))\n",
"\n",
@@ -450,7 +619,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 5,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:08:26.640466Z",
@@ -473,14 +642,22 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 6,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:08:26.645227Z",
"start_time": "2020-05-22T14:08:26.642010Z"
}
},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Input shape = (347227, 89), target shape = (347227,)\n"
+ ]
+ }
+ ],
"source": [
"print(\"Input shape = {}, target shape = {}\".format(X_ElemNet.shape, y_vol_per_atom.shape))"
]
@@ -501,7 +678,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 7,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:08:26.793126Z",
@@ -529,14 +706,25 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 8,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:08:26.799383Z",
"start_time": "2020-05-22T14:08:26.795789Z"
}
},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "'\\nscaler = StandardScaler()\\nX = scaler.fit_transform(X)\\nX_test = scaler.transform(X_test)\\n'"
+ ]
+ },
+ "execution_count": 8,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
"source": [
"\"\"\"\n",
"scaler = StandardScaler()\n",
@@ -554,7 +742,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 9,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:08:26.921060Z",
@@ -590,7 +778,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 10,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:08:26.928103Z",
@@ -663,14 +851,66 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 11,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:08:27.092710Z",
"start_time": "2020-05-22T14:08:26.930206Z"
}
},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Model: \"model_1\"\n",
+ "_________________________________________________________________\n",
+ "Layer (type) Output Shape Param # \n",
+ "=================================================================\n",
+ "x_input (InputLayer) (None, 89) 0 \n",
+ "_________________________________________________________________\n",
+ "dense_1 (Dense) (None, 512) 46080 \n",
+ "_________________________________________________________________\n",
+ "dropout_1 (Dropout) (None, 512) 0 \n",
+ "_________________________________________________________________\n",
+ "dense_2 (Dense) (None, 256) 131328 \n",
+ "_________________________________________________________________\n",
+ "dropout_2 (Dropout) (None, 256) 0 \n",
+ "_________________________________________________________________\n",
+ "dense_3 (Dense) (None, 128) 32896 \n",
+ "_________________________________________________________________\n",
+ "dropout_3 (Dropout) (None, 128) 0 \n",
+ "_________________________________________________________________\n",
+ "dense_4 (Dense) (None, 64) 8256 \n",
+ "_________________________________________________________________\n",
+ "dropout_4 (Dropout) (None, 64) 0 \n",
+ "_________________________________________________________________\n",
+ "dense_5 (Dense) (None, 32) 2080 \n",
+ "_________________________________________________________________\n",
+ "dropout_5 (Dropout) (None, 32) 0 \n",
+ "_________________________________________________________________\n",
+ "dense_6 (Dense) (None, 18) 594 \n",
+ "_________________________________________________________________\n",
+ "dropout_6 (Dropout) (None, 18) 0 \n",
+ "_________________________________________________________________\n",
+ "dense_7 (Dense) (None, 8) 152 \n",
+ "_________________________________________________________________\n",
+ "dropout_7 (Dropout) (None, 8) 0 \n",
+ "_________________________________________________________________\n",
+ "dense_8 (Dense) (None, 4) 36 \n",
+ "_________________________________________________________________\n",
+ "dropout_8 (Dropout) (None, 4) 0 \n",
+ "_________________________________________________________________\n",
+ "dense_9 (Dense) (None, 1) 5 \n",
+ "=================================================================\n",
+ "Total params: 221,427\n",
+ "Trainable params: 221,427\n",
+ "Non-trainable params: 0\n",
+ "_________________________________________________________________\n",
+ "None\n"
+ ]
+ }
+ ],
"source": [
"batch_size = 64\n",
"epochs = 30\n",
@@ -698,7 +938,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 12,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:15:47.803232Z",
@@ -706,7 +946,75 @@
},
"scrolled": true
},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Train on 222224 samples, validate on 55557 samples\n",
+ "Epoch 1/30\n",
+ "222224/222224 [==============================] - 11s 52us/step - loss: 12.1375 - val_loss: 6.3870\n",
+ "Epoch 2/30\n",
+ "222224/222224 [==============================] - 11s 49us/step - loss: 6.3106 - val_loss: 6.1132\n",
+ "Epoch 3/30\n",
+ "222224/222224 [==============================] - 11s 49us/step - loss: 5.9615 - val_loss: 5.2210\n",
+ "Epoch 4/30\n",
+ "222224/222224 [==============================] - 12s 55us/step - loss: 5.6623 - val_loss: 5.3325\n",
+ "Epoch 5/30\n",
+ "222224/222224 [==============================] - 12s 54us/step - loss: 5.5168 - val_loss: 5.2384\n",
+ "Epoch 6/30\n",
+ "222224/222224 [==============================] - 11s 50us/step - loss: 5.3398 - val_loss: 6.1497\n",
+ "Epoch 7/30\n",
+ "222224/222224 [==============================] - 11s 50us/step - loss: 5.2586 - val_loss: 4.9812\n",
+ "Epoch 8/30\n",
+ "222224/222224 [==============================] - 12s 54us/step - loss: 5.1311 - val_loss: 5.3434\n",
+ "Epoch 9/30\n",
+ "222224/222224 [==============================] - 12s 53us/step - loss: 5.0557 - val_loss: 4.9242\n",
+ "Epoch 10/30\n",
+ "222224/222224 [==============================] - 12s 53us/step - loss: 4.9974 - val_loss: 5.0442\n",
+ "Epoch 11/30\n",
+ "222224/222224 [==============================] - 12s 53us/step - loss: 4.9424 - val_loss: 4.9540\n",
+ "Epoch 12/30\n",
+ "222224/222224 [==============================] - 12s 54us/step - loss: 4.8965 - val_loss: 5.1100\n",
+ "Epoch 13/30\n",
+ "222224/222224 [==============================] - 11s 51us/step - loss: 4.8423 - val_loss: 5.0179\n",
+ "Epoch 14/30\n",
+ "222224/222224 [==============================] - 11s 48us/step - loss: 4.7907 - val_loss: 4.7615\n",
+ "Epoch 15/30\n",
+ "222224/222224 [==============================] - 11s 51us/step - loss: 4.7590 - val_loss: 4.8152\n",
+ "Epoch 16/30\n",
+ "222224/222224 [==============================] - 11s 49us/step - loss: 4.7547 - val_loss: 4.7692\n",
+ "Epoch 17/30\n",
+ "222224/222224 [==============================] - 11s 48us/step - loss: 4.6649 - val_loss: 5.0429\n",
+ "Epoch 18/30\n",
+ "222224/222224 [==============================] - 12s 56us/step - loss: 4.6749 - val_loss: 4.7545\n",
+ "Epoch 19/30\n",
+ "222224/222224 [==============================] - 11s 50us/step - loss: 4.6553 - val_loss: 4.7230\n",
+ "Epoch 20/30\n",
+ "222224/222224 [==============================] - 11s 51us/step - loss: 4.6322 - val_loss: 4.7478\n",
+ "Epoch 21/30\n",
+ "222224/222224 [==============================] - 12s 52us/step - loss: 4.5951 - val_loss: 4.7244\n",
+ "Epoch 22/30\n",
+ "222224/222224 [==============================] - 12s 55us/step - loss: 4.5834 - val_loss: 4.7551\n",
+ "Epoch 23/30\n",
+ "222224/222224 [==============================] - 11s 52us/step - loss: 4.5497 - val_loss: 4.6306\n",
+ "Epoch 24/30\n",
+ "222224/222224 [==============================] - 13s 59us/step - loss: 4.5465 - val_loss: 4.7262\n",
+ "Epoch 25/30\n",
+ "222224/222224 [==============================] - 13s 56us/step - loss: 4.5242 - val_loss: 4.6449\n",
+ "Epoch 26/30\n",
+ "222224/222224 [==============================] - 13s 57us/step - loss: 4.5048 - val_loss: 4.7496\n",
+ "Epoch 27/30\n",
+ "222224/222224 [==============================] - 14s 62us/step - loss: 4.4913 - val_loss: 4.5988\n",
+ "Epoch 28/30\n",
+ "222224/222224 [==============================] - 15s 67us/step - loss: 4.4824 - val_loss: 4.7221\n",
+ "Epoch 29/30\n",
+ "222224/222224 [==============================] - 13s 59us/step - loss: 4.4582 - val_loss: 4.6347\n",
+ "Epoch 30/30\n",
+ "222224/222224 [==============================] - 13s 58us/step - loss: 4.4414 - val_loss: 4.8915\n"
+ ]
+ }
+ ],
"source": [
"history = model.fit(X_train, y_train,\n",
" validation_data = (X_val, y_val),\n",
@@ -723,14 +1031,27 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 13,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:15:47.945244Z",
"start_time": "2020-05-22T14:15:47.805420Z"
}
},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "<Figure size 432x288 with 1 Axes>"
+ ]
+ },
+ "metadata": {
+ "needs_background": "light"
+ },
+ "output_type": "display_data"
+ }
+ ],
"source": [
"import matplotlib.pyplot as plt\n",
"\n",
@@ -768,14 +1089,27 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 14,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:15:53.924363Z",
"start_time": "2020-05-22T14:15:47.947273Z"
}
},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Training: MAE: 1.011 | MSE: 4.305 | RMSE: 2.075 | Pearson: 0.972\n",
+ "Dataset: Min: 2.726 | Max: 98.717 | Mean: 22.149 | Std.dev.: 8.588\n",
+ "\n",
+ "\n",
+ "Validation: MAE: 1.070 | MSE: 4.892 | RMSE: 2.212 | Pearson: 0.969\n",
+ "Dataset: Min: 2.724 | Max: 99.106 | Mean: 22.167 | Std.dev.: 8.672\n"
+ ]
+ }
+ ],
"source": [
"def rmse(y_pred, y_true):\n",
" return np.sqrt(((y_pred - y_true) ** 2).mean())\n",
@@ -830,14 +1164,25 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 15,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:15:56.919219Z",
"start_time": "2020-05-22T14:15:53.926422Z"
}
},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "<Figure size 432x432 with 3 Axes>"
+ ]
+ },
+ "metadata": {},
+ "output_type": "display_data"
+ }
+ ],
"source": [
"y_true = y_true_val\n",
"y_pred = y_pred_val\n",
@@ -877,7 +1222,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 16,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:15:56.923919Z",
@@ -891,7 +1236,7 @@
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 17,
"metadata": {
"ExecuteTime": {
"end_time": "2020-05-22T14:16:02.573238Z",
@@ -899,7 +1244,25 @@
},
"scrolled": true
},
- "outputs": [],
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "Test: MAE: 1.082 | MSE: 4.913 | RMSE: 2.217 | Pearson: 0.968\n"
+ ]
+ },
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "<Figure size 432x432 with 3 Axes>"
+ ]
+ },
+ "metadata": {},
+ "output_type": "display_data"
+ }
+ ],
"source": [
"if investigate_test_set:\n",
" y_pred_test = model.predict(X_test).flatten()\n",
--
GitLab