quickstart.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### The training and evaluation of a cell-based NARVAL R2B4 model\n",
    "\n",
    "This notebook is a quickly illustrates the process of training and evaluating a cell-based neural network based cloud cover parameterization.<br> We train and evaluate the neural network on coarse-grained and preprocessed NARVAL R2B4 data. In this notebook, we work with data that is already prepared for this purpose."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU'),\n",
       " PhysicalDevice(name='/physical_device:GPU:1', device_type='GPU'),\n",
       " PhysicalDevice(name='/physical_device:GPU:2', device_type='GPU'),\n",
       " PhysicalDevice(name='/physical_device:GPU:3', device_type='GPU')]"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Importing necessary packages\n",
    "import sys\n",
    "import time\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import matplotlib.pyplot as plt\n",
    "import os\n",
    "\n",
    "from tensorflow.keras.models import load_model\n",
    "from tensorflow.keras.optimizers import Nadam\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense\n",
    "\n",
    "# Print options for pandas\n",
    "pd.options.display.float_format = \"{:,.2g}\".format\n",
    "\n",
    "# Reloading custom file to incorporate changes dynamically\n",
    "import importlib\n",
    "import my_classes\n",
    "importlib.reload(my_classes)\n",
    "\n",
    "# Data path\n",
    "pwd = os.path.abspath('.')\n",
    "path_data = pwd + '/extract_from_the_data/coarse-grained_and_preprocessed/narval_r2b4_cell_based'\n",
    "\n",
    "# Stop training after how many minutes.\n",
    "# We decreased the training time for this notebook to an hour.\n",
    "from my_classes import TimeOut\n",
    "t0 = time.time()\n",
    "timeout = 60 \n",
    "\n",
    "# Random seed\n",
    "NUM = 1 \n",
    "tf.random.set_seed(NUM)\n",
    "\n",
    "# List GPUs\n",
    "gpus = tf.config.experimental.list_physical_devices('GPU')\n",
    "gpus"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Loading the data\n",
    "**Input-variables:**\n",
    "- *fr_land*: Fraction of land \n",
    "- *z_g*: Geometric height at full levels\n",
    "- *q_v*: Specific water vapor content\n",
    "- *q_i*: Specific cloud ice content\n",
    "- *temp*: Temperature\n",
    "- *pres*: Pressure\n",
    "\n",
    "**Output-variables:**\n",
    "- *clc*: Cloud cover (cloud volume fraction)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load coarse-grained and preprocessed data. I split up the data so that it fits into the 100MB file size limit on GitHub\n",
    "input_train = [np.load(path_data + '/cloud_cover_all_days_input_train_%d.npy'%k) for k in range(14)]\n",
    "input_valid = [np.load(path_data + '/cloud_cover_all_days_input_valid_%d.npy'%k) for k in range(2)]\n",
    "input_test = [np.load(path_data + '/cloud_cover_all_days_input_test_%d.npy'%k) for k in range(4)]\n",
    "output_train = [np.load(path_data + '/cloud_cover_all_days_output_train_%d.npy'%k) for k in range(3)]\n",
    "\n",
    "input_train = np.concatenate(input_train, axis=0)\n",
    "input_valid = np.concatenate(input_valid, axis=0)\n",
    "input_test = np.concatenate(input_test, axis=0)\n",
    "output_train = np.concatenate(output_train, axis=0)\n",
    "output_valid = np.load(path_data + '/cloud_cover_all_days_output_valid.npy')\n",
    "output_test = np.load(path_data + '/cloud_cover_all_days_output_test.npy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The training data consists of 26482169 training samples.\n",
      "The validation data consists of 2942463 validation samples.\n",
      "The test data consists of 7356158 test samples.\n",
      "We have 6 input features for the neural network.\n"
     ]
    }
   ],
   "source": [
    "print('The training data consists of %d training samples.'%input_train.shape[0])\n",
    "print('The validation data consists of %d validation samples.'%input_valid.shape[0])\n",
    "print('The test data consists of %d test samples.'%input_test.shape[0])\n",
    "\n",
    "print('We have %d input features for the neural network.'%input_train.shape[1])\n",
    "assert input_train.shape[1] == input_valid.shape[1] == input_test.shape[1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Looking at the data\n",
    "\n",
    "The corresponding means, standard deviations and features can be found in: <br>\n",
    "/n1_cell_based_narval_r2b4/saved_models/model_grid_cell_based_v3_final_1.txt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "features = ['qv', 'qi', 'temp', 'pres', 'zg', 'fr_land']\n",
    "mean = np.array([5.37518440e-03, 4.65389731e-07, 2.59635412e+02, 5.52329389e+04, 6.79260772e+03, 2.58097095e-01])\n",
    "std = np.array([6.01943993e-03, 3.95009930e-06, 3.55940285e+01, 3.26642242e+04, 6.20726361e+03, 4.28313535e-01])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "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>qv</th>\n",
       "      <th>qi</th>\n",
       "      <th>temp</th>\n",
       "      <th>pres</th>\n",
       "      <th>zg</th>\n",
       "      <th>fr_land</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>2.6e+07</td>\n",
       "      <td>2.6e+07</td>\n",
       "      <td>2.6e+07</td>\n",
       "      <td>2.6e+07</td>\n",
       "      <td>2.6e+07</td>\n",
       "      <td>2.6e+07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>2e-16</td>\n",
       "      <td>1.1e-16</td>\n",
       "      <td>-1.2e-14</td>\n",
       "      <td>3e-15</td>\n",
       "      <td>1.5e-16</td>\n",
       "      <td>1.1e-15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>-0.89</td>\n",
       "      <td>-0.12</td>\n",
       "      <td>-2.1</td>\n",
       "      <td>-1.5</td>\n",
       "      <td>-1.1</td>\n",
       "      <td>-0.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>-0.88</td>\n",
       "      <td>-0.12</td>\n",
       "      <td>-0.86</td>\n",
       "      <td>-0.94</td>\n",
       "      <td>-0.88</td>\n",
       "      <td>-0.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>-0.5</td>\n",
       "      <td>-0.12</td>\n",
       "      <td>0.38</td>\n",
       "      <td>0.11</td>\n",
       "      <td>-0.35</td>\n",
       "      <td>-0.6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.95</td>\n",
       "      <td>-0.12</td>\n",
       "      <td>0.87</td>\n",
       "      <td>0.98</td>\n",
       "      <td>0.68</td>\n",
       "      <td>1.3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>2.5</td>\n",
       "      <td>2.4e+02</td>\n",
       "      <td>1.6</td>\n",
       "      <td>1.4</td>\n",
       "      <td>2.3</td>\n",
       "      <td>1.7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           qv      qi     temp    pres      zg  fr_land\n",
       "count 2.6e+07 2.6e+07  2.6e+07 2.6e+07 2.6e+07  2.6e+07\n",
       "mean    2e-16 1.1e-16 -1.2e-14   3e-15 1.5e-16  1.1e-15\n",
       "std         1       1        1       1       1        1\n",
       "min     -0.89   -0.12     -2.1    -1.5    -1.1     -0.6\n",
       "25%     -0.88   -0.12    -0.86   -0.94   -0.88     -0.6\n",
       "50%      -0.5   -0.12     0.38    0.11   -0.35     -0.6\n",
       "75%      0.95   -0.12     0.87    0.98    0.68      1.3\n",
       "max       2.5 2.4e+02      1.6     1.4     2.3      1.7"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Looking at the data. It is already normalized.\n",
    "pd.DataFrame(input_train, columns=features).describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# The validation/test data is normalized w.r.t. the training data, so e.g. the means are also close to zero\n",
    "assert 1e-3 < np.sum(np.abs(input_valid.mean(axis=0))) < 1e-2\n",
    "assert 1e-3 < np.sum(np.abs(input_test.mean(axis=0))) < 1e-2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "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>qv</th>\n",
       "      <th>qi</th>\n",
       "      <th>temp</th>\n",
       "      <th>pres</th>\n",
       "      <th>zg</th>\n",
       "      <th>fr_land</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>2.6e+07</td>\n",
       "      <td>2.6e+07</td>\n",
       "      <td>2.6e+07</td>\n",
       "      <td>2.6e+07</td>\n",
       "      <td>2.6e+07</td>\n",
       "      <td>2.6e+07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.0054</td>\n",
       "      <td>4.7e-07</td>\n",
       "      <td>2.6e+02</td>\n",
       "      <td>5.5e+04</td>\n",
       "      <td>6.8e+03</td>\n",
       "      <td>0.26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.006</td>\n",
       "      <td>4e-06</td>\n",
       "      <td>36</td>\n",
       "      <td>3.3e+04</td>\n",
       "      <td>6.2e+03</td>\n",
       "      <td>0.43</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>1.2e-06</td>\n",
       "      <td>-1.3e-16</td>\n",
       "      <td>1.9e+02</td>\n",
       "      <td>4.8e+03</td>\n",
       "      <td>20</td>\n",
       "      <td>1.1e-10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>6e-05</td>\n",
       "      <td>-1.3e-16</td>\n",
       "      <td>2.3e+02</td>\n",
       "      <td>2.5e+04</td>\n",
       "      <td>1.3e+03</td>\n",
       "      <td>1.1e-10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.0024</td>\n",
       "      <td>-1.2e-16</td>\n",
       "      <td>2.7e+02</td>\n",
       "      <td>5.9e+04</td>\n",
       "      <td>4.6e+03</td>\n",
       "      <td>1.1e-10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.011</td>\n",
       "      <td>6e-16</td>\n",
       "      <td>2.9e+02</td>\n",
       "      <td>8.7e+04</td>\n",
       "      <td>1.1e+04</td>\n",
       "      <td>0.82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>0.021</td>\n",
       "      <td>0.00095</td>\n",
       "      <td>3.2e+02</td>\n",
       "      <td>1e+05</td>\n",
       "      <td>2.1e+04</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           qv       qi    temp    pres      zg  fr_land\n",
       "count 2.6e+07  2.6e+07 2.6e+07 2.6e+07 2.6e+07  2.6e+07\n",
       "mean   0.0054  4.7e-07 2.6e+02 5.5e+04 6.8e+03     0.26\n",
       "std     0.006    4e-06      36 3.3e+04 6.2e+03     0.43\n",
       "min   1.2e-06 -1.3e-16 1.9e+02 4.8e+03      20  1.1e-10\n",
       "25%     6e-05 -1.3e-16 2.3e+02 2.5e+04 1.3e+03  1.1e-10\n",
       "50%    0.0024 -1.2e-16 2.7e+02 5.9e+04 4.6e+03  1.1e-10\n",
       "75%     0.011    6e-16 2.9e+02 8.7e+04 1.1e+04     0.82\n",
       "max     0.021  0.00095 3.2e+02   1e+05 2.1e+04        1"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Reverse scaling to look at the original data\n",
    "pd.DataFrame((input_train*std)+mean, columns=features).describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### The neural network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 256)               1792      \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 256)               65792     \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 257       \n",
      "=================================================================\n",
      "Total params: 67,841\n",
      "Trainable params: 67,841\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Set up the neural network\n",
    "model = Sequential()\n",
    "model.add(Dense(256, activation='relu', input_dim = 6))\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dense(1, activation='linear'))\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Here, we train the neural network**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Starting training\n",
      "Epoch 1/70\n",
      "51723/51723 [==============================] - 243s 5ms/step - loss: 27.5949 - val_loss: 18.3962\n",
      "Epoch 2/70\n",
      "51723/51723 [==============================] - 238s 5ms/step - loss: 18.3481 - val_loss: 18.2304\n",
      "Epoch 3/70\n",
      "51723/51723 [==============================] - 241s 5ms/step - loss: 17.6708 - val_loss: 17.8188\n",
      "Epoch 4/70\n",
      "51723/51723 [==============================] - 248s 5ms/step - loss: 17.3494 - val_loss: 17.4409\n",
      "Epoch 5/70\n",
      "51723/51723 [==============================] - 237s 5ms/step - loss: 17.1172 - val_loss: 16.7895\n",
      "Epoch 6/70\n",
      "51723/51723 [==============================] - 242s 5ms/step - loss: 16.9198 - val_loss: 16.5442\n",
      "Epoch 7/70\n",
      "51723/51723 [==============================] - 241s 5ms/step - loss: 16.7319 - val_loss: 16.9072\n",
      "Epoch 8/70\n",
      "51723/51723 [==============================] - 243s 5ms/step - loss: 16.5933 - val_loss: 16.2358\n",
      "Epoch 9/70\n",
      "51723/51723 [==============================] - 241s 5ms/step - loss: 16.4655 - val_loss: 16.3494\n",
      "Epoch 10/70\n",
      "51723/51723 [==============================] - 233s 5ms/step - loss: 16.3324 - val_loss: 16.0755\n",
      "Epoch 11/70\n",
      "51723/51723 [==============================] - 247s 5ms/step - loss: 16.2916 - val_loss: 16.0633\n",
      "Epoch 12/70\n",
      "51723/51723 [==============================] - 242s 5ms/step - loss: 16.1737 - val_loss: 16.6643\n",
      "Epoch 13/70\n",
      "51723/51723 [==============================] - 231s 4ms/step - loss: 16.1244 - val_loss: 16.1331\n",
      "Epoch 14/70\n",
      "51723/51723 [==============================] - 231s 4ms/step - loss: 16.0196 - val_loss: 16.0688\n",
      "Epoch 15/70\n",
      "48524/51723 [===========================>..] - ETA: 13s - loss: 15.9902\n",
      "Reached 60.000 minutes of training, stopping\n",
      "51723/51723 [==============================] - 217s 4ms/step - loss: 15.9882 - val_loss: 16.1479\n",
      "Restore model weights from the end of the best epoch\n"
     ]
    }
   ],
   "source": [
    "model.compile(loss='mse', optimizer=Nadam())\n",
    "# Incorporates early stopping, i.e. the weights from the best concluded epoch are used\n",
    "time_callback = TimeOut(t0, timeout) \n",
    "# We increase the batch size from 32 (in the paper) to 512 to require much less time per epoch\n",
    "history = model.fit(input_train, output_train, batch_size=512, epochs=70, verbose=1,\n",
    "                    validation_data=(input_valid, output_valid), callbacks=[time_callback])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Let's look at how the training proceeded**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Plotting the training progress\n",
    "if len(history.history['loss']) > len(history.history['val_loss']):\n",
    "    del history.history['loss'][-1]\n",
    "pd.DataFrame(history.history).plot(figsize=(8,5))\n",
    "plt.grid(True)\n",
    "plt.ylabel('Mean Squared Error')\n",
    "plt.xlabel('Number of epochs')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**We evaluate the mean-squared errors on the training/validation/test sets** <br>\n",
    "In the paper, we have MSEs of around 15 for the cell-based NARVAL R2B4 model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "13242/13242 - 23s - loss: 15.9891\n",
      "2943/2943 - 4s - loss: 16.0633\n",
      "7357/7357 - 11s - loss: 16.0345\n",
      "\n",
      "Training loss: 15.9891\n",
      "Validation loss: 16.0633\n",
      "Test loss: 16.0345\n",
      "Training epochs: 15\n"
     ]
    }
   ],
   "source": [
    "train_loss = model.evaluate(input_train, output_train, verbose=2, batch_size=2000)\n",
    "valid_loss = model.evaluate(input_valid, output_valid, verbose=2, batch_size=1000)\n",
    "test_loss = model.evaluate(input_test, output_test, verbose=2, batch_size=1000)\n",
    "\n",
    "print('\\nTraining loss: %.4f'%(train_loss))\n",
    "print('Validation loss: %.4f'%(valid_loss))\n",
    "print('Test loss: %.4f'%(test_loss))\n",
    "print('Training epochs: %d'%(len(history.history['val_loss'])))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Evaluation of the neural network\n",
    "\n",
    "In the data we have just worked with (input_train, input_valid, input_test), where we do not know the vertical layer of the data samples. <br> We cannot use that data to evaluate the neural networks using vertical profiles in this example. <br> What we thus did was to load the NARVAL R2B4 data and to store it as an npy file (which saves space) in /extract_from_the_data/narval_r2b4_data_for_the_quickstart.npy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # We ran this cell once:\n",
    "\n",
    "# from my_classes import load_data\n",
    "\n",
    "# # Loads the NARVAL data into the data_dict dictionary\n",
    "# order_of_vars=['qv', 'qi', 'temp', 'pres', 'zg', 'fr_land', 'clc']\n",
    "# data_dict = load_data(source='narval', days='all', vert_interp=True, order_of_vars=order_of_vars)\n",
    "\n",
    "# #Reshaping into nd-arrays of equaling shapes (have timesteps x vert x hor)\n",
    "# data_dict['zg'] = np.repeat(np.expand_dims(data_dict['zg'], 0), data_dict['qv'].shape[0], axis=0)\n",
    "# data_dict['fr_land'] = np.repeat(np.expand_dims(data_dict['fr_land'], 0), data_dict['qv'].shape[0], axis=0)\n",
    "# data_dict['fr_land'] = np.repeat(np.expand_dims(data_dict['fr_land'], 1), data_dict['qv'].shape[1], axis=1)\n",
    "\n",
    "# assert data_dict['fr_land'].shape == data_dict['qv'].shape == data_dict['zg'].shape\n",
    "\n",
    "# # Convert data_dict into numpy file and save as npy. We cannot take more than 54 time steps due to size limits on GitHub.\n",
    "# np_arr = np.array([data_dict[key][:54] for key in data_dict.keys()])\n",
    "# np.save('./extract_from_the_data/narval_r2b4_data_for_the_quickstart.npy', np_arr)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**First evaluation method** <br>\n",
    "Samples on x-axis and cloud cover on y-axis. This we can do without knowing which vertical layer corresponds to a data sample."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2b39f07a1f90>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "no_samples = 50\n",
    "clc_pred = model.predict(input_test[:no_samples])[:,0]\n",
    "clc_true = output_test[:no_samples]\n",
    "a = np.linspace(1, no_samples, no_samples)\n",
    "\n",
    "fig = plt.figure(figsize=(15,4))\n",
    "ax = fig.add_subplot(111, title='NNs predictions on random samples from the test set', \n",
    "                     xlabel='Sample', ylabel='Cloud Cover')\n",
    "ax.plot(a, clc_pred, 'ro')\n",
    "ax.plot(a, clc_true, 'bo')\n",
    "ax.legend(['Predictions', 'Truth'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Second evaluation method** <br>\n",
    "Mean profile of Cloud Cover on x-axis with Vertical Layer on y-axis <br>\n",
    "Here, I'm taking only the first 54 time steps of the NARVAL R2B4 data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(7, 54, 27, 1024)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load and scale the data\n",
    "np_arr = np.load('./extract_from_the_data/narval_r2b4_data_for_the_quickstart.npy')\n",
    "scaled_data = np.array([(np_arr[i] - mean[i])/std[i] for i in range(len(features))])\n",
    "scaled_data = np.append(scaled_data, np_arr[-1:], axis=0) # Add cloud cover\n",
    "# Removes data above 21 km\n",
    "scaled_data = scaled_data[:,:,4:,:]\n",
    "scaled_data.shape # features x time steps x vertical layers x horizontal fields"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Predicted cloud cover mean per vertical layer\n",
    "clc_pred_mean = []\n",
    "for v_layer in range(21, 48):\n",
    "    pred = np.transpose(model.predict_on_batch(np.transpose(scaled_data[:-1,:,v_layer-21,:])))\n",
    "    pred_adj = np.minimum(np.maximum(pred, 0), 100)\n",
    "    clc_pred_mean.append(np.mean(pred_adj))\n",
    "    \n",
    "# Cloud cover mean in the data\n",
    "clc_data_mean = np.mean(scaled_data[-1], axis=(0, 2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2b94dc2eaa90>"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot mean cloud cover per vertical layer\n",
    "a = np.linspace(21, 48, 27)\n",
    "fig = plt.figure(figsize=(7,4))\n",
    "ax = fig.add_subplot(111, xlabel='Mean Cloud Cover [%]', ylabel='Vertical layer')\n",
    "ax.plot(clc_data_mean, a)\n",
    "ax.plot(clc_pred_mean, a)\n",
    "plt.gca().invert_yaxis()\n",
    "ax.legend(['Ground truth', 'Prediction'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2b94dc33dd50>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "zg_mean = np.mean(np_arr[np.where([features[i] == 'zg' for i in range(len(features))])[0][0], :, 4:, :], axis=(0,2))\n",
    "\n",
    "# Plot mean cloud cover per averaged height\n",
    "fig = plt.figure(figsize=(7,4))\n",
    "ax = fig.add_subplot(111, xlabel='Mean Cloud Cover [%]', ylabel='z [km]')\n",
    "ax.plot(clc_data_mean, zg_mean)\n",
    "ax.plot(clc_pred_mean, zg_mean)\n",
    "ax.legend(['Ground truth', 'Prediction'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Third visualization method** <br>\n",
    "R2-value on x-axis with vertical layers on y-axis <br>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "# r2 = 1-MSE/VAR\n",
    "r2 = []\n",
    "for v_layer in range(23, 48):\n",
    "    pred = np.transpose(model.predict_on_batch(np.transpose(scaled_data[:-1,:,v_layer-21,:])))\n",
    "    pred_adj = np.minimum(np.maximum(pred, 0), 100)\n",
    "    \n",
    "    mse = np.mean((pred_adj - scaled_data[-1, :, v_layer-21, :])**2)\n",
    "    var = np.var(scaled_data[-1, :, v_layer-21, :])\n",
    "    r2.append(1-mse/var)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2b94dc4cc510>]"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "# Plotting\n",
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(111, xlim=(0, 1), title='R2-values of the cell-based model excluding the two upper-most layers',\n",
    "                     xlabel='R2-value/Coefficient of determination', ylabel='z [km]')\n",
    "ax.plot(r2, zg_mean[2:], 'bo', ls='--')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "clouds",
   "language": "python",
   "name": "clouds"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}