Forecasting_Daily_Minimum_Temperatures_Melbourne.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Week 4: Using real world data\n",
    "\n",
    "Welcome! So far you have worked exclusively with generated data. This time you will be using the [Daily Minimum Temperatures in Melbourne](https://github.com/jbrownlee/Datasets/blob/master/daily-min-temperatures.csv) dataset which contains data of the daily minimum temperatures recorded in Melbourne from 1981 to 1990. In addition to be using Tensorflow's layers for processing sequence data such as Recurrent layers or LSTMs you will also use Convolutional layers to improve the model's performance.\n",
    "\n",
    "Let's get started!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "_**NOTE:** To prevent errors from the autograder, you are not allowed to edit or delete some of the cells in this notebook . Please only put your solutions in between the `### START CODE HERE` and `### END CODE HERE` code comments, and also refrain from adding any new cells. **Once you have passed this assignment** and want to experiment with any of the locked cells, you may follow the instructions at the bottom of this notebook._"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "deletable": false,
    "editable": false,
    "id": "56XEQOGknrAk",
    "outputId": "6b8ab5ea-ed49-40d2-a27c-d2ba08710ba0"
   },
   "outputs": [],
   "source": [
    "import csv\n",
    "import pickle\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import matplotlib.pyplot as plt\n",
    "from dataclasses import dataclass\n",
    "from absl import logging\n",
    "logging.set_verbosity(logging.ERROR)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Begin by looking at the structure of the csv that contains the data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "deletable": false,
    "editable": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Header looks like this:\n",
      "\n",
      "\"Date\",\"Temp\"\n",
      "\n",
      "First data point looks like this:\n",
      "\n",
      "\"1981-01-01\",20.7\n",
      "\n",
      "Second data point looks like this:\n",
      "\n",
      "\"1981-01-02\",17.9\n",
      "\n"
     ]
    }
   ],
   "source": [
    "TEMPERATURES_CSV = './data/daily-min-temperatures.csv'\n",
    "\n",
    "with open(TEMPERATURES_CSV, 'r') as csvfile:\n",
    "    print(f\"Header looks like this:\\n\\n{csvfile.readline()}\")    \n",
    "    print(f\"First data point looks like this:\\n\\n{csvfile.readline()}\")\n",
    "    print(f\"Second data point looks like this:\\n\\n{csvfile.readline()}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you can see, each data point is composed of the date and the recorded minimum temperature for that date.\n",
    "\n",
    "\n",
    "In the first exercise you will code a function to read the data from the csv but for now run the next cell to load a helper function to plot the time series."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "deletable": false,
    "editable": false,
    "id": "sLl52leVp5wU"
   },
   "outputs": [],
   "source": [
    "def plot_series(time, series, format=\"-\", start=0, end=None):\n",
    "    plt.plot(time[start:end], series[start:end], format)\n",
    "    plt.xlabel(\"Time\")\n",
    "    plt.ylabel(\"Value\")\n",
    "    plt.grid(True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Parsing the raw data\n",
    "Now you need to read the data from the csv file. To do so, complete the `parse_data_from_file` function.\n",
    "\n",
    "A couple of things to note:\n",
    "\n",
    "- You should omit the first line as the file contains headers.\n",
    "- There is no need to save the data points as numpy arrays, regular lists is fine.\n",
    "- To read from csv files use `csv.reader` by passing the appropriate arguments.\n",
    "- `csv.reader` returns an iterable that returns each row in every iteration. So the temperature can be accessed via row[1] and the date can be discarded.\n",
    "- The `times` list should contain every timestep (starting at zero), which is just a sequence of ordered numbers with the same length as the `temperatures` list.\n",
    "- The values of the `temperatures` should be of `float` type. You can use Python's built-in `float` function to ensure this."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 388
    },
    "deletable": false,
    "id": "NcG9r1eClbTh",
    "outputId": "7acf6ba9-e852-4f06-e06e-b0ff1b9e2ddd"
   },
   "outputs": [],
   "source": [
    "def parse_data_from_file(filename):\n",
    "    \n",
    "    times = []\n",
    "    temperatures = []\n",
    "\n",
    "    with open(filename) as csvfile:\n",
    "        \n",
    "        ### START CODE HERE\n",
    "        \n",
    "        reader = csv.reader(csvfile, delimiter=',')\n",
    "            \n",
    "        # Skip the first line\n",
    "        next(reader)\n",
    "        \n",
    "        # Append row and sunspot number to lists\n",
    "        for index, row in enumerate(reader):\n",
    "            temperatures.append(float(row[1]))  # Convert temperature to float\n",
    "            times.append(index)  # Create a sequential list starting from 0\n",
    "        ### END CODE HERE\n",
    "            \n",
    "    return times, temperatures"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The next cell will use your function to compute the `times` and `temperatures` and will save these as numpy arrays within the `G` dataclass. This cell will also plot the time series:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "deletable": false,
    "editable": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Test your function and save all \"global\" variables within the G class (G stands for global)\n",
    "@dataclass\n",
    "class G:\n",
    "    TEMPERATURES_CSV = './data/daily-min-temperatures.csv'\n",
    "    times, temperatures = parse_data_from_file(TEMPERATURES_CSV)\n",
    "    TIME = np.array(times)\n",
    "    SERIES = np.array(temperatures)\n",
    "    SPLIT_TIME = 2500\n",
    "    WINDOW_SIZE = 64\n",
    "    BATCH_SIZE = 32\n",
    "    SHUFFLE_BUFFER_SIZE = 1000\n",
    "\n",
    "\n",
    "plt.figure(figsize=(10, 6))\n",
    "plot_series(G.TIME, G.SERIES)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Expected Output:**\n",
    "<div>\n",
    "<img src=\"images/temp-series.png\" width=\"500\"/>\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Processing the data\n",
    "\n",
    "Since you already coded the `train_val_split` and `windowed_dataset` functions during past week's assignments, this time they are provided for you:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "deletable": false,
    "editable": false
   },
   "outputs": [],
   "source": [
    "def train_val_split(time, series, time_step=G.SPLIT_TIME):\n",
    "\n",
    "    time_train = time[:time_step]\n",
    "    series_train = series[:time_step]\n",
    "    time_valid = time[time_step:]\n",
    "    series_valid = series[time_step:]\n",
    "\n",
    "    return time_train, series_train, time_valid, series_valid\n",
    "\n",
    "\n",
    "# Split the dataset\n",
    "time_train, series_train, time_valid, series_valid = train_val_split(G.TIME, G.SERIES)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "deletable": false,
    "editable": false,
    "id": "lJwUUZscnG38"
   },
   "outputs": [],
   "source": [
    "def windowed_dataset(series, window_size=G.WINDOW_SIZE, batch_size=G.BATCH_SIZE, shuffle_buffer=G.SHUFFLE_BUFFER_SIZE):\n",
    "    ds = tf.data.Dataset.from_tensor_slices(series)\n",
    "    ds = ds.window(window_size + 1, shift=1, drop_remainder=True)\n",
    "    ds = ds.flat_map(lambda w: w.batch(window_size + 1))\n",
    "    ds = ds.shuffle(shuffle_buffer)\n",
    "    ds = ds.map(lambda w: (w[:-1], w[-1]))\n",
    "    ds = ds.batch(batch_size).prefetch(1)\n",
    "    return ds\n",
    "\n",
    "\n",
    "# Apply the transformation to the training set\n",
    "train_set = windowed_dataset(series_train, window_size=G.WINDOW_SIZE, batch_size=G.BATCH_SIZE, shuffle_buffer=G.SHUFFLE_BUFFER_SIZE)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Defining the model architecture\n",
    "\n",
    "Now that you have a function that will process the data before it is fed into your neural network for training, it is time to define your layer architecture. Just as in last week's assignment you will do the layer definition and compilation in two separate steps. Begin by completing the `create_uncompiled_model` function below.\n",
    "\n",
    "This is done so you can reuse your model's layers for the learning rate adjusting and the actual training.\n",
    "\n",
    "Hint:\n",
    "\n",
    "- `Lambda` layers are not required.\n",
    "- Use a combination of `Conv1D` and `LSTM` layers followed by `Dense` layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "deletable": false
   },
   "outputs": [],
   "source": [
    "def create_uncompiled_model():\n",
    "\n",
    "    ### START CODE HERE\n",
    "    \n",
    "    model = tf.keras.models.Sequential([\n",
    "        tf.keras.layers.Conv1D(filters=64, kernel_size=5, activation=\"relu\", padding='causal', input_shape=[G.WINDOW_SIZE, 1]),\n",
    "        tf.keras.layers.LSTM(64, return_sequences=True),\n",
    "        tf.keras.layers.LSTM(64),\n",
    "        tf.keras.layers.Dense(30, activation=\"relu\"),\n",
    "        tf.keras.layers.Dense(10, activation=\"relu\"),\n",
    "        tf.keras.layers.Dense(1)\n",
    "    ]) \n",
    "    \n",
    "    ### END CODE HERE\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can test your model with the code below. If you get an error, it's likely that your model is returning a sequence. You can indeed use an LSTM with `return_sequences=True` but you have to feed it into another layer that generates a single prediction. You can review the lectures or the previous ungraded labs to see how that is done."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "deletable": false,
    "editable": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Testing model prediction with input of shape (32, 64)...\n",
      "Your current architecture is compatible with the windowed dataset! :)\n"
     ]
    }
   ],
   "source": [
    "# Test your uncompiled model\n",
    "\n",
    "# Create an instance of the model\n",
    "uncompiled_model = create_uncompiled_model()\n",
    "\n",
    "# Get one batch of the training set(X = input, y = label)\n",
    "for X, y in train_set.take(1):\n",
    "    \n",
    "    # Generate a prediction\n",
    "    print(f'Testing model prediction with input of shape {X.shape}...')\n",
    "    y_pred = uncompiled_model.predict(X)\n",
    "    \n",
    "# Compare the shape of the prediction and the label y (remove dimensions of size 1)\n",
    "y_pred_shape = y_pred.squeeze().shape\n",
    "\n",
    "assert y_pred_shape == y.shape, (f'Squeezed predicted y shape = {y_pred_shape} '\n",
    "                                           f'whereas actual y shape = {y.shape}.')\n",
    "\n",
    "print(\"Your current architecture is compatible with the windowed dataset! :)\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Adjusting the learning rate - (Optional Exercise)\n",
    "\n",
    "As you saw in the lectures, you can leverage Tensorflow's callbacks to dynamically vary the learning rate before doing the actual training. This can be helpful in finding what value works best with your model. Note that this is only one way of finding the best learning rate. There are other techniques for hyperparameter optimization but it is outside the scope of this course.\n",
    "\n",
    "For the optimizers you can try out:\n",
    "\n",
    "- tf.keras.optimizers.Adam\n",
    "- tf.keras.optimizers.SGD with a momentum of 0.9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "deletable": false
   },
   "outputs": [],
   "source": [
    "def adjust_learning_rate(dataset):\n",
    "    \n",
    "    model = create_uncompiled_model()\n",
    "    \n",
    "    lr_schedule = tf.keras.callbacks.LearningRateScheduler(lambda epoch: 1e-5 * 10**(epoch / 10))\n",
    "    \n",
    "    ### START CODE HERE\n",
    "    \n",
    "    # Select your optimizer\n",
    "    optimizer = tf.keras.optimizers.Adam(learning_rate=1e-4)\n",
    "    \n",
    "    # Compile the model passing in the appropriate loss\n",
    "    model.compile(loss=tf.keras.losses.Huber(),\n",
    "                  optimizer=optimizer, \n",
    "                  metrics=[\"mae\"]) \n",
    "    \n",
    "    ### END CODE HERE\n",
    "    \n",
    "    history = model.fit(dataset, epochs=100, callbacks=[lr_schedule])\n",
    "    \n",
    "    return history"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "deletable": false,
    "editable": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "77/77 [==============================] - 12s 113ms/step - loss: 10.1276 - mae: 10.6264 - lr: 1.0000e-05\n",
      "Epoch 2/100\n",
      "77/77 [==============================] - 8s 108ms/step - loss: 9.8422 - mae: 10.3408 - lr: 1.2589e-05\n",
      "Epoch 3/100\n",
      "77/77 [==============================] - 9s 110ms/step - loss: 9.2884 - mae: 9.7865 - lr: 1.5849e-05\n",
      "Epoch 4/100\n",
      "77/77 [==============================] - 9s 115ms/step - loss: 8.4528 - mae: 8.9501 - lr: 1.9953e-05\n",
      "Epoch 5/100\n",
      "77/77 [==============================] - 9s 112ms/step - loss: 7.1349 - mae: 7.6307 - lr: 2.5119e-05\n",
      "Epoch 6/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 5.6561 - mae: 6.1436 - lr: 3.1623e-05\n",
      "Epoch 7/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 4.3645 - mae: 4.8452 - lr: 3.9811e-05\n",
      "Epoch 8/100\n",
      "77/77 [==============================] - 8s 103ms/step - loss: 3.3673 - mae: 3.8399 - lr: 5.0119e-05\n",
      "Epoch 9/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 2.8491 - mae: 3.3163 - lr: 6.3096e-05\n",
      "Epoch 10/100\n",
      "77/77 [==============================] - 8s 107ms/step - loss: 2.4336 - mae: 2.8935 - lr: 7.9433e-05\n",
      "Epoch 11/100\n",
      "77/77 [==============================] - 8s 103ms/step - loss: 2.0238 - mae: 2.4766 - lr: 1.0000e-04\n",
      "Epoch 12/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 1.8707 - mae: 2.3195 - lr: 1.2589e-04\n",
      "Epoch 13/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 1.8179 - mae: 2.2680 - lr: 1.5849e-04\n",
      "Epoch 14/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 1.8010 - mae: 2.2539 - lr: 1.9953e-04\n",
      "Epoch 15/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 1.7776 - mae: 2.2258 - lr: 2.5119e-04\n",
      "Epoch 16/100\n",
      "77/77 [==============================] - 8s 104ms/step - loss: 1.8128 - mae: 2.2664 - lr: 3.1623e-04\n",
      "Epoch 17/100\n",
      "77/77 [==============================] - 8s 103ms/step - loss: 1.7136 - mae: 2.1644 - lr: 3.9811e-04\n",
      "Epoch 18/100\n",
      "77/77 [==============================] - 8s 105ms/step - loss: 1.6559 - mae: 2.1071 - lr: 5.0119e-04\n",
      "Epoch 19/100\n",
      "77/77 [==============================] - 8s 109ms/step - loss: 1.6040 - mae: 2.0467 - lr: 6.3096e-04\n",
      "Epoch 20/100\n",
      "77/77 [==============================] - 8s 104ms/step - loss: 1.5814 - mae: 2.0245 - lr: 7.9433e-04\n",
      "Epoch 21/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 1.6041 - mae: 2.0520 - lr: 0.0010\n",
      "Epoch 22/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 1.5982 - mae: 2.0440 - lr: 0.0013\n",
      "Epoch 23/100\n",
      "77/77 [==============================] - 8s 106ms/step - loss: 1.5506 - mae: 1.9901 - lr: 0.0016\n",
      "Epoch 24/100\n",
      "77/77 [==============================] - 8s 107ms/step - loss: 1.5854 - mae: 2.0297 - lr: 0.0020\n",
      "Epoch 25/100\n",
      "77/77 [==============================] - 8s 105ms/step - loss: 1.6488 - mae: 2.0968 - lr: 0.0025\n",
      "Epoch 26/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 1.5821 - mae: 2.0261 - lr: 0.0032\n",
      "Epoch 27/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 1.6700 - mae: 2.1165 - lr: 0.0040\n",
      "Epoch 28/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 1.6221 - mae: 2.0709 - lr: 0.0050\n",
      "Epoch 29/100\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 1.6382 - mae: 2.0825 - lr: 0.0063\n",
      "Epoch 30/100\n",
      "77/77 [==============================] - 8s 98ms/step - loss: 1.6416 - mae: 2.0867 - lr: 0.0079\n",
      "Epoch 31/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 1.6888 - mae: 2.1355 - lr: 0.0100\n",
      "Epoch 32/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 1.8511 - mae: 2.3033 - lr: 0.0126\n",
      "Epoch 33/100\n",
      "77/77 [==============================] - 8s 99ms/step - loss: 1.7204 - mae: 2.1736 - lr: 0.0158\n",
      "Epoch 34/100\n",
      "77/77 [==============================] - 8s 98ms/step - loss: 1.7697 - mae: 2.2205 - lr: 0.0200\n",
      "Epoch 35/100\n",
      "77/77 [==============================] - 8s 104ms/step - loss: 1.7637 - mae: 2.2156 - lr: 0.0251\n",
      "Epoch 36/100\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 1.7904 - mae: 2.2438 - lr: 0.0316\n",
      "Epoch 37/100\n",
      "77/77 [==============================] - 8s 99ms/step - loss: 1.8187 - mae: 2.2709 - lr: 0.0398\n",
      "Epoch 38/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 2.0144 - mae: 2.4705 - lr: 0.0501\n",
      "Epoch 39/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 1.8853 - mae: 2.3406 - lr: 0.0631\n",
      "Epoch 40/100\n",
      "77/77 [==============================] - 8s 103ms/step - loss: 2.1613 - mae: 2.6195 - lr: 0.0794\n",
      "Epoch 41/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 2.3248 - mae: 2.7848 - lr: 0.1000\n",
      "Epoch 42/100\n",
      "77/77 [==============================] - 8s 98ms/step - loss: 3.1131 - mae: 3.5832 - lr: 0.1259\n",
      "Epoch 43/100\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 2.8330 - mae: 3.3011 - lr: 0.1585\n",
      "Epoch 44/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 2.7722 - mae: 3.2398 - lr: 0.1995\n",
      "Epoch 45/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 2.7720 - mae: 3.2406 - lr: 0.2512\n",
      "Epoch 46/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 2.7702 - mae: 3.2387 - lr: 0.3162\n",
      "Epoch 47/100\n",
      "77/77 [==============================] - 8s 99ms/step - loss: 2.7555 - mae: 3.2200 - lr: 0.3981\n",
      "Epoch 48/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 2.7814 - mae: 3.2496 - lr: 0.5012\n",
      "Epoch 49/100\n",
      "77/77 [==============================] - 8s 103ms/step - loss: 2.7895 - mae: 3.2570 - lr: 0.6310\n",
      "Epoch 50/100\n",
      "77/77 [==============================] - 8s 104ms/step - loss: 2.7276 - mae: 3.1929 - lr: 0.7943\n",
      "Epoch 51/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 2.7374 - mae: 3.2050 - lr: 1.0000\n",
      "Epoch 52/100\n",
      "77/77 [==============================] - 8s 103ms/step - loss: 2.7316 - mae: 3.1981 - lr: 1.2589\n",
      "Epoch 53/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 2.7283 - mae: 3.1947 - lr: 1.5849\n",
      "Epoch 54/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 2.7365 - mae: 3.2024 - lr: 1.9953\n",
      "Epoch 55/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 2.7933 - mae: 3.2606 - lr: 2.5119\n",
      "Epoch 56/100\n",
      "77/77 [==============================] - 8s 104ms/step - loss: 2.8050 - mae: 3.2708 - lr: 3.1623\n",
      "Epoch 57/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 2.8569 - mae: 3.3228 - lr: 3.9811\n",
      "Epoch 58/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 2.8684 - mae: 3.3335 - lr: 5.0119\n",
      "Epoch 59/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 2.8822 - mae: 3.3542 - lr: 6.3096\n",
      "Epoch 60/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 2.9313 - mae: 3.3972 - lr: 7.9433\n",
      "Epoch 61/100\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 3.0553 - mae: 3.5257 - lr: 10.0000\n",
      "Epoch 62/100\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 3.1259 - mae: 3.5953 - lr: 12.5893\n",
      "Epoch 63/100\n",
      "77/77 [==============================] - 8s 99ms/step - loss: 2.9614 - mae: 3.4314 - lr: 15.8489\n",
      "Epoch 64/100\n",
      "77/77 [==============================] - 8s 103ms/step - loss: 3.4303 - mae: 3.9032 - lr: 19.9526\n",
      "Epoch 65/100\n",
      "77/77 [==============================] - 8s 99ms/step - loss: 3.4823 - mae: 3.9573 - lr: 25.1189\n",
      "Epoch 66/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 3.4629 - mae: 3.9372 - lr: 31.6228\n",
      "Epoch 67/100\n",
      "77/77 [==============================] - 8s 108ms/step - loss: 4.1494 - mae: 4.6254 - lr: 39.8107\n",
      "Epoch 68/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 5.0790 - mae: 5.5625 - lr: 50.1187\n",
      "Epoch 69/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 4.3195 - mae: 4.7977 - lr: 63.0957\n",
      "Epoch 70/100\n",
      "77/77 [==============================] - 8s 98ms/step - loss: 7.2534 - mae: 7.7428 - lr: 79.4328\n",
      "Epoch 71/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 9.4433 - mae: 9.9417 - lr: 100.0000\n",
      "Epoch 72/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 14.8182 - mae: 15.3141 - lr: 125.8925\n",
      "Epoch 73/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 10.5105 - mae: 10.9987 - lr: 158.4893\n",
      "Epoch 74/100\n",
      "77/77 [==============================] - 8s 98ms/step - loss: 24.4370 - mae: 24.9336 - lr: 199.5262\n",
      "Epoch 75/100\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 36.3046 - mae: 36.8046 - lr: 251.1886\n",
      "Epoch 76/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 44.1543 - mae: 44.6543 - lr: 316.2278\n",
      "Epoch 77/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 52.8030 - mae: 53.3030 - lr: 398.1072\n",
      "Epoch 78/100\n",
      "77/77 [==============================] - 8s 98ms/step - loss: 63.9941 - mae: 64.4941 - lr: 501.1872\n",
      "Epoch 79/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 79.3246 - mae: 79.8246 - lr: 630.9573\n",
      "Epoch 80/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 96.2604 - mae: 96.7604 - lr: 794.3282\n",
      "Epoch 81/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 118.6526 - mae: 119.1526 - lr: 1000.0000\n",
      "Epoch 82/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 148.1343 - mae: 148.6343 - lr: 1258.9254\n",
      "Epoch 83/100\n",
      "77/77 [==============================] - 8s 103ms/step - loss: 180.7926 - mae: 181.2926 - lr: 1584.8932\n",
      "Epoch 84/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 224.2032 - mae: 224.7032 - lr: 1995.2623\n",
      "Epoch 85/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 281.5868 - mae: 282.0868 - lr: 2511.8865\n",
      "Epoch 86/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 345.3543 - mae: 345.8543 - lr: 3162.2776\n",
      "Epoch 87/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 430.4322 - mae: 430.9322 - lr: 3981.0718\n",
      "Epoch 88/100\n",
      "77/77 [==============================] - 8s 99ms/step - loss: 542.1323 - mae: 542.6323 - lr: 5011.8726\n",
      "Epoch 89/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 667.2912 - mae: 667.7912 - lr: 6309.5732\n",
      "Epoch 90/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 833.9373 - mae: 834.4373 - lr: 7943.2822\n",
      "Epoch 91/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 1053.5157 - mae: 1054.0157 - lr: 10000.0000\n",
      "Epoch 92/100\n",
      "77/77 [==============================] - 8s 99ms/step - loss: 1299.3433 - mae: 1299.8433 - lr: 12589.2539\n",
      "Epoch 93/100\n",
      "77/77 [==============================] - 8s 98ms/step - loss: 1627.4805 - mae: 1627.9805 - lr: 15848.9316\n",
      "Epoch 94/100\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 2060.2751 - mae: 2060.7751 - lr: 19952.6230\n",
      "Epoch 95/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 2545.3391 - mae: 2545.8391 - lr: 25118.8652\n",
      "Epoch 96/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 3192.7866 - mae: 3193.2866 - lr: 31622.7773\n",
      "Epoch 97/100\n",
      "77/77 [==============================] - 8s 99ms/step - loss: 4047.3796 - mae: 4047.8796 - lr: 39810.7188\n",
      "Epoch 98/100\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 5006.8511 - mae: 5007.3511 - lr: 50118.7227\n",
      "Epoch 99/100\n",
      "77/77 [==============================] - 8s 98ms/step - loss: 6287.9858 - mae: 6288.4858 - lr: 63095.7344\n",
      "Epoch 100/100\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 7979.3491 - mae: 7979.8491 - lr: 79432.8203\n"
     ]
    }
   ],
   "source": [
    "# Run the training with dynamic LR\n",
    "lr_history = adjust_learning_rate(train_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 290
    },
    "deletable": false,
    "editable": false,
    "id": "vVcKmg7Q_7rD",
    "outputId": "27cf16ae-eb85-47c3-fc86-18e72e528619"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.0001, 10.0, 0.0, 10.0)"
      ]
     },
     "execution_count": 50,
     "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": [
    "plt.semilogx(lr_history.history[\"lr\"], lr_history.history[\"loss\"])\n",
    "plt.axis([1e-4, 10, 0, 10])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Compiling the model\n",
    "Now that you have trained the model while varying the learning rate, it is time to do the actual training that will be used to forecast the time series. For this complete the `create_model` function below.\n",
    "\n",
    "Notice that you are reusing the architecture you defined in the `create_uncompiled_model` earlier. Now you only need to compile this model using the appropriate loss, optimizer (and learning rate).\n",
    "\n",
    "Hints:\n",
    "\n",
    "- The training should be really quick so if you notice that each epoch is taking more than a few seconds, consider trying a different architecture.\n",
    "\n",
    "\n",
    "- If after the first epoch you get an output like this: loss: nan - mae: nan it is very likely that your network is suffering from exploding gradients. This is a common problem if you used SGD as optimizer and set a learning rate that is too high. If you encounter this problem consider lowering the learning rate or using Adam with the default learning rate."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "deletable": false
   },
   "outputs": [],
   "source": [
    "def create_model():\n",
    "\n",
    "    \n",
    "    model = create_uncompiled_model()\n",
    "\n",
    "    ### START CODE HERE\n",
    "\n",
    "    optimizer = tf.keras.optimizers.Adam(learning_rate=1e-4)\n",
    "        \n",
    "    model.compile(loss=tf.keras.losses.Huber(),\n",
    "                  optimizer=optimizer, \n",
    "                  metrics=[\"mae\"])  \n",
    "    \n",
    "\n",
    "    ### END CODE HERE\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "deletable": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "77/77 [==============================] - 10s 97ms/step - loss: 9.0320 - mae: 9.5297\n",
      "Epoch 2/50\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 5.8621 - mae: 6.3520\n",
      "Epoch 3/50\n",
      "77/77 [==============================] - 7s 96ms/step - loss: 3.3500 - mae: 3.8215\n",
      "Epoch 4/50\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 2.5384 - mae: 3.0009\n",
      "Epoch 5/50\n",
      "77/77 [==============================] - 7s 96ms/step - loss: 2.0730 - mae: 2.5292\n",
      "Epoch 6/50\n",
      "77/77 [==============================] - 7s 96ms/step - loss: 1.8999 - mae: 2.3510\n",
      "Epoch 7/50\n",
      "77/77 [==============================] - 7s 96ms/step - loss: 1.8393 - mae: 2.2892\n",
      "Epoch 8/50\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 1.8143 - mae: 2.2631\n",
      "Epoch 9/50\n",
      "77/77 [==============================] - 7s 95ms/step - loss: 1.8097 - mae: 2.2591\n",
      "Epoch 10/50\n",
      "77/77 [==============================] - 7s 95ms/step - loss: 1.7955 - mae: 2.2455\n",
      "Epoch 11/50\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 1.7858 - mae: 2.2340\n",
      "Epoch 12/50\n",
      "77/77 [==============================] - 8s 98ms/step - loss: 1.7609 - mae: 2.2074\n",
      "Epoch 13/50\n",
      "77/77 [==============================] - 7s 96ms/step - loss: 1.7530 - mae: 2.2038\n",
      "Epoch 14/50\n",
      "77/77 [==============================] - 8s 99ms/step - loss: 1.7368 - mae: 2.1851\n",
      "Epoch 15/50\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 1.7090 - mae: 2.1576\n",
      "Epoch 16/50\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 1.7065 - mae: 2.1559\n",
      "Epoch 17/50\n",
      "77/77 [==============================] - 8s 98ms/step - loss: 1.6997 - mae: 2.1511\n",
      "Epoch 18/50\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 1.6744 - mae: 2.1236\n",
      "Epoch 19/50\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 1.6656 - mae: 2.1139\n",
      "Epoch 20/50\n",
      "77/77 [==============================] - 8s 102ms/step - loss: 1.6541 - mae: 2.1035\n",
      "Epoch 21/50\n",
      "77/77 [==============================] - 8s 99ms/step - loss: 1.6224 - mae: 2.0685\n",
      "Epoch 22/50\n",
      "77/77 [==============================] - 8s 98ms/step - loss: 1.6094 - mae: 2.0574\n",
      "Epoch 23/50\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 1.5887 - mae: 2.0368\n",
      "Epoch 24/50\n",
      "77/77 [==============================] - 8s 99ms/step - loss: 1.5736 - mae: 2.0177\n",
      "Epoch 25/50\n",
      "77/77 [==============================] - 7s 95ms/step - loss: 1.5615 - mae: 2.0066\n",
      "Epoch 26/50\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 1.5695 - mae: 2.0132\n",
      "Epoch 27/50\n",
      "77/77 [==============================] - 7s 94ms/step - loss: 1.5458 - mae: 1.9892\n",
      "Epoch 28/50\n",
      "77/77 [==============================] - 7s 91ms/step - loss: 1.5297 - mae: 1.9722\n",
      "Epoch 29/50\n",
      "77/77 [==============================] - 7s 95ms/step - loss: 1.5213 - mae: 1.9618\n",
      "Epoch 30/50\n",
      "77/77 [==============================] - 7s 91ms/step - loss: 1.5098 - mae: 1.9508\n",
      "Epoch 31/50\n",
      "77/77 [==============================] - 7s 94ms/step - loss: 1.4945 - mae: 1.9339\n",
      "Epoch 32/50\n",
      "77/77 [==============================] - 7s 94ms/step - loss: 1.5122 - mae: 1.9514\n",
      "Epoch 33/50\n",
      "77/77 [==============================] - 7s 94ms/step - loss: 1.5021 - mae: 1.9420\n",
      "Epoch 34/50\n",
      "77/77 [==============================] - 7s 95ms/step - loss: 1.4901 - mae: 1.9288\n",
      "Epoch 35/50\n",
      "77/77 [==============================] - 7s 93ms/step - loss: 1.4900 - mae: 1.9319\n",
      "Epoch 36/50\n",
      "77/77 [==============================] - 7s 93ms/step - loss: 1.4933 - mae: 1.9331\n",
      "Epoch 37/50\n",
      "77/77 [==============================] - 7s 93ms/step - loss: 1.5059 - mae: 1.9459\n",
      "Epoch 38/50\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 1.4938 - mae: 1.9344\n",
      "Epoch 39/50\n",
      "77/77 [==============================] - 8s 98ms/step - loss: 1.4846 - mae: 1.9235\n",
      "Epoch 40/50\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 1.4800 - mae: 1.9178\n",
      "Epoch 41/50\n",
      "77/77 [==============================] - 7s 95ms/step - loss: 1.4750 - mae: 1.9136\n",
      "Epoch 42/50\n",
      "77/77 [==============================] - 7s 93ms/step - loss: 1.4696 - mae: 1.9088\n",
      "Epoch 43/50\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 1.4754 - mae: 1.9153\n",
      "Epoch 44/50\n",
      "77/77 [==============================] - 8s 103ms/step - loss: 1.4672 - mae: 1.9064\n",
      "Epoch 45/50\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 1.4768 - mae: 1.9133\n",
      "Epoch 46/50\n",
      "77/77 [==============================] - 8s 101ms/step - loss: 1.4921 - mae: 1.9332\n",
      "Epoch 47/50\n",
      "77/77 [==============================] - 8s 104ms/step - loss: 1.4759 - mae: 1.9131\n",
      "Epoch 48/50\n",
      "77/77 [==============================] - 8s 100ms/step - loss: 1.4574 - mae: 1.8938\n",
      "Epoch 49/50\n",
      "77/77 [==============================] - 8s 105ms/step - loss: 1.4697 - mae: 1.9097\n",
      "Epoch 50/50\n",
      "77/77 [==============================] - 8s 97ms/step - loss: 1.4648 - mae: 1.9019\n"
     ]
    }
   ],
   "source": [
    "# Save an instance of the model\n",
    "model = create_model()\n",
    "\n",
    "# Train it\n",
    "history = model.fit(train_set, epochs=50)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluating the forecast\n",
    "\n",
    "Now it is time to evaluate the performance of the forecast. For this you can use the `compute_metrics` function that you coded in a previous assignment:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "deletable": false,
    "editable": false
   },
   "outputs": [],
   "source": [
    "def compute_metrics(true_series, forecast):\n",
    "    \n",
    "    mse = tf.keras.metrics.mean_squared_error(true_series, forecast).numpy()\n",
    "    mae = tf.keras.metrics.mean_absolute_error(true_series, forecast).numpy()\n",
    "\n",
    "    return mse, mae"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "At this point only the model that will perform the forecast is ready but you still need to compute the actual forecast.\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Faster model forecasts\n",
    "\n",
    "In the previous week you saw a faster approach compared to using a for loop to compute the forecasts for every point in the sequence. Remember that this faster approach uses batches of data. \n",
    "\n",
    "The code to implement this is provided in the `model_forecast` below. Notice that the code is very similar to the one in the `windowed_dataset` function with the differences that:\n",
    "- The dataset is windowed using `window_size` rather than `window_size + 1`\n",
    "- No shuffle should be used\n",
    "- No need to split the data into features and labels\n",
    "- A model is used to predict batches of the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "deletable": false,
    "editable": false,
    "id": "4XwGrf-A_wF0"
   },
   "outputs": [],
   "source": [
    "def model_forecast(model, series, window_size):\n",
    "    ds = tf.data.Dataset.from_tensor_slices(series)\n",
    "    ds = ds.window(window_size, shift=1, drop_remainder=True)\n",
    "    ds = ds.flat_map(lambda w: w.batch(window_size))\n",
    "    ds = ds.batch(32).prefetch(1)\n",
    "    forecast = model.predict(ds)\n",
    "    return forecast"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now compute the actual forecast:\n",
    "\n",
    "**Note:** Don't modify the cell below. \n",
    "\n",
    "The grader uses the same slicing to get the forecast so if you change the cell below you risk having issues when submitting your model for grading."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 389
    },
    "deletable": false,
    "editable": false,
    "id": "PrktQX3hKYex",
    "outputId": "1914662d-6bdd-4e17-8697-8f5a29e89b87"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:5 out of the last 117 calls to <function Model.make_predict_function.<locals>.predict_function at 0x7472d4460b80> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Compute the forecast for all the series\n",
    "rnn_forecast = model_forecast(model, G.SERIES, G.WINDOW_SIZE).squeeze()\n",
    "\n",
    "# Slice the forecast to get only the predictions for the validation set\n",
    "rnn_forecast = rnn_forecast[G.SPLIT_TIME - G.WINDOW_SIZE:-1]\n",
    "\n",
    "# Plot the forecast\n",
    "plt.figure(figsize=(10, 6))\n",
    "plot_series(time_valid, series_valid)\n",
    "plot_series(time_valid, rnn_forecast)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "deletable": false,
    "editable": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mse: 5.34, mae: 1.80 for forecast\n"
     ]
    }
   ],
   "source": [
    "mse, mae = compute_metrics(series_valid, rnn_forecast)\n",
    "\n",
    "print(f\"mse: {mse:.2f}, mae: {mae:.2f} for forecast\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**To pass this assignment your forecast should achieve a MSE of 6 or less and a MAE of 2 or less.**\n",
    "\n",
    "- If your forecast didn't achieve this threshold try re-training your model with a different architecture (you will need to re-run both `create_uncompiled_model` and `create_model` functions) or tweaking the optimizer's parameters.\n",
    "\n",
    "\n",
    "- If your forecast did achieve this threshold run the following cell to save the model in the SavedModel format which will be used for grading and after doing so, submit your assigment for grading.\n",
    "\n",
    "\n",
    "- This environment includes a dummy SavedModel directory which contains a dummy model trained for one epoch. **To replace this file with your actual model you need to run the next cell before submitting for grading.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "deletable": false,
    "editable": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: saved_model/my_model/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: saved_model/my_model/assets\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "saved_model/\n",
      "saved_model/my_model/\n",
      "saved_model/my_model/keras_metadata.pb\n",
      "saved_model/my_model/variables/\n",
      "saved_model/my_model/variables/variables.data-00000-of-00001\n",
      "saved_model/my_model/variables/variables.index\n",
      "saved_model/my_model/saved_model.pb\n",
      "saved_model/my_model/assets/\n"
     ]
    }
   ],
   "source": [
    "# Save your model in the SavedModel format\n",
    "model.save('saved_model/my_model')\n",
    "\n",
    "# Compress the directory using tar\n",
    "! tar -czvf saved_model.tar.gz saved_model/"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Congratulations on finishing this week's assignment!**\n",
    "\n",
    "You have successfully implemented a neural network capable of forecasting time series leveraging a combination of Tensorflow's layers such as Convolutional and LSTMs! This resulted in a forecast that surpasses all the ones you did previously.\n",
    "\n",
    "**By finishing this assignment you have finished the specialization! Give yourself a pat on the back!!!**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<details>\n",
    "  <summary><font size=\"2\" color=\"darkgreen\"><b>Please click here if you want to experiment with any of the non-graded code.</b></font></summary>\n",
    "    <p><i><b>Important Note: Please only do this when you've already passed the assignment to avoid problems with the autograder.</b></i>\n",
    "    <ol>\n",
    "        <li> On the notebook’s menu, click “View” > “Cell Toolbar” > “Edit Metadata”</li>\n",
    "        <li> Hit the “Edit Metadata” button next to the code cell which you want to lock/unlock</li>\n",
    "        <li> Set the attribute value for “editable” to:\n",
    "            <ul>\n",
    "                <li> “true” if you want to unlock it </li>\n",
    "                <li> “false” if you want to lock it </li>\n",
    "            </ul>\n",
    "        </li>\n",
    "        <li> On the notebook’s menu, click “View” > “Cell Toolbar” > “None” </li>\n",
    "    </ol>\n",
    "    <p> Here's a short demo of how to do the steps above: \n",
    "        <br>\n",
    "        <img src=\"https://drive.google.com/uc?export=view&id=14Xy_Mb17CZVgzVAgq7NCjMVBvSae3xO1\" align=\"center\">\n",
    "</details>"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}