Iris_Classification.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Iris_Classification.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "metadata": {
        "id": "3K6owcJtB23q",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "25eafc75-e56a-46ff-a7bc-db472accde22"
      },
      "cell_type": "code",
      "source": [
        "# The Iris classification problem\n",
        "# https://github.com/tensorflow/models/blob/master/samples/core/tutorials/eager/custom_training_walkthrough.ipynb\n",
        "\n",
        "# Setup the Program\n",
        "# Configure imports and eager execution\n",
        "# Trained in Google Colab\n",
        "\n",
        "from __future__ import absolute_import, division, print_function\n",
        "\n",
        "import os\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "import tensorflow as tf\n",
        "import tensorflow.contrib.eager as tfe\n",
        "\n",
        "tf.enable_eager_execution()\n",
        "\n",
        "print(\"TensorFlow Version: {}\".format(tf.VERSION))\n",
        "print(\"Eager Execution: {}\".format(tf.executing_eagerly()))\n"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "TensorFlow Version: 1.10.0-rc1\n",
            "Eager Execution: True\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "No-TV0aYDxNe",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "4e7c8200-e76e-4b3a-a914-b64055f22c13"
      },
      "cell_type": "code",
      "source": [
        "# Import and parse the training dataset\n",
        "## Download the dataset file and convert it into a structure that can be used by this Python program.\n",
        "\n",
        "# Download the dataset\n",
        "## Download the training dataset file using the tf.keras.utils.get_file function. This returns the file path of the downloaded file.\n",
        "\n",
        "train_dataset_url = \"http://download.tensorflow.org/data/iris_training.csv\"\n",
        "\n",
        "train_dataset_file_path = tf.keras.utils.get_file(fname = os.path.basename(train_dataset_url), origin = train_dataset_url)\n",
        "\n",
        "print(\"Local copy of dataset: {}\".format(train_dataset_url))"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Local copy of dataset: http://download.tensorflow.org/data/iris_training.csv\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "tGPGUE5aGXb4",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 102
        },
        "outputId": "2f730668-3c86-4f49-a079-edbdb549a51d"
      },
      "cell_type": "code",
      "source": [
        "# Inspect the data\n",
        "# Use the head -n5 command to take a peak at the first five entries\n",
        "\n",
        "!head -n5 {train_dataset_file_path}\n"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "120,4,setosa,versicolor,virginica\r\n",
            "6.4,2.8,5.6,2.2,2\r\n",
            "5.0,2.3,3.3,1.0,1\r\n",
            "4.9,2.5,4.5,1.7,2\r\n",
            "4.9,3.1,1.5,0.1,0\r\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "u-tHgNKmG-ZM",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "87a99f8f-e82a-4105-9f68-e5a5925f631c"
      },
      "cell_type": "code",
      "source": [
        "# column order in CSV file\n",
        "column_names = ['sepal_length', 'sepal_width', 'petal_length', 'petal_width', 'species']\n",
        "feature_names = column_names[:-1]\n",
        "label_names = column_names[-1]\n",
        "\n",
        "print(\"feature names: {}\".format(feature_names))\n",
        "print(\"label names: {}\".format(label_names))"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "feature names: ['sepal_length', 'sepal_width', 'petal_length', 'petal_width']\n",
            "label names: species\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "31CjsA-OE6J8",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# 0: Iris setosa\n",
        "# 1: Iris versicolor\n",
        "# 2: Iris virginica\n",
        "class_name = ['Iris setosa', 'Iris versicolor', 'Iris virginica']\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "4y_aEL9JFn6h",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Create a tf.data.Dataset\n",
        "\n",
        "batch_size = 32\n",
        "\n",
        "train_dataset = tf.contrib.data.make_csv_dataset(train_dataset_file_path, \n",
        "                                      batch_size, \n",
        "                                      label_name = label_names, \n",
        "                                      column_names = column_names, \n",
        "                                      num_epochs = 1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "Hf8xL0TtIfIh",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 357
        },
        "outputId": "5c32573f-51bc-4146-d331-0950f4a54294"
      },
      "cell_type": "code",
      "source": [
        "# With eager execution enabled, these Dataset objects are iterable.\n",
        "features, labels = next(iter(train_dataset))\n",
        "features\n"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "OrderedDict([('sepal_length',\n",
              "              <tf.Tensor: id=60, shape=(32,), dtype=float32, numpy=\n",
              "              array([5.6, 4.9, 5.8, 5.1, 4.6, 5.7, 5.4, 5. , 6. , 6.9, 5.4, 6. , 5.5,\n",
              "                     5.9, 6. , 6.7, 7.2, 4.9, 6.4, 5. , 5.8, 6.3, 6.5, 4.8, 6.3, 4.6,\n",
              "                     5.4, 6.3, 6.1, 5.8, 6.7, 6. ], dtype=float32)>),\n",
              "             ('sepal_width',\n",
              "              <tf.Tensor: id=61, shape=(32,), dtype=float32, numpy=\n",
              "              array([2.7, 2.4, 2.7, 3.8, 3.6, 2.8, 3. , 3.2, 2.2, 3.1, 3.9, 3. , 2.4,\n",
              "                     3. , 2.9, 3.1, 3.2, 2.5, 2.8, 3.5, 2.7, 2.7, 2.8, 3.4, 2.5, 3.1,\n",
              "                     3.9, 2.3, 2.8, 2.7, 3.1, 2.7], dtype=float32)>),\n",
              "             ('petal_length',\n",
              "              <tf.Tensor: id=58, shape=(32,), dtype=float32, numpy=\n",
              "              array([4.2, 3.3, 5.1, 1.6, 1. , 4.1, 4.5, 1.2, 5. , 5.1, 1.3, 4.8, 3.8,\n",
              "                     5.1, 4.5, 4.4, 6. , 4.5, 5.6, 1.6, 5.1, 4.9, 4.6, 1.6, 5. , 1.5,\n",
              "                     1.7, 4.4, 4. , 4.1, 5.6, 5.1], dtype=float32)>),\n",
              "             ('petal_width',\n",
              "              <tf.Tensor: id=59, shape=(32,), dtype=float32, numpy=\n",
              "              array([1.3, 1. , 1.9, 0.2, 0.2, 1.3, 1.5, 0.2, 1.5, 2.3, 0.4, 1.8, 1.1,\n",
              "                     1.8, 1.5, 1.4, 1.8, 1.7, 2.2, 0.6, 1.9, 1.8, 1.5, 0.2, 1.9, 0.2,\n",
              "                     0.4, 1.3, 1.3, 1. , 2.4, 1.6], dtype=float32)>)])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "metadata": {
        "id": "jA2lIX5QJHDu",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 378
        },
        "outputId": "6ceea37f-430c-4c13-c4d8-2fb7fe4768e1"
      },
      "cell_type": "code",
      "source": [
        "# clusters by plotting a few features from the batch\n",
        "plt.scatter(features['petal_length'], \n",
        "           features['sepal_length'], \n",
        "           c = labels,\n",
        "           cmap='viridis')\n",
        "\n",
        "plt.xlabel('petal_length')\n",
        "plt.ylabel('sepal_length')"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0,0.5,'sepal_length')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFYCAYAAAB6RnQAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7fe743dcfa90>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "qKo97A4bJ68F",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# create a function to repackage the features dictionary into a single array with shape: (batch_size, num_features)\n",
        "# This function uses the tf.stack method which takes values from a list of tensors and creates a combined tensor at the specified dimension.\n",
        "def pack_feature_vector(features, labels):\n",
        "  # pack the features into a single array\n",
        "  features = tf.stack(list(features.values()), axis = 1)\n",
        "  return features, labels"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "v308y3kRKmw0",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Then use the tf.data.Dataset.map method to pack the features of each (features,label) pair into the training dataset\n",
        "train_dataset = train_dataset.map(pack_feature_vector)\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "0o1ZP8TjLJcq",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        },
        "outputId": "44610d5d-fc15-493f-f69f-b6b7e5a8709a"
      },
      "cell_type": "code",
      "source": [
        "# check the results\n",
        "features, labels = next(iter(train_dataset))\n",
        "print(features[:5])"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "tf.Tensor(\n",
            "[[5.6 2.9 3.6 1.3]\n",
            " [6.1 2.9 4.7 1.4]\n",
            " [4.5 2.3 1.3 0.3]\n",
            " [7.  3.2 4.7 1.4]\n",
            " [5.1 3.8 1.9 0.4]], shape=(5, 4), dtype=float32)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "ehLfiV6pLZbk",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Create a model using Keras\n",
        "model = tf.keras.Sequential([\n",
        "    tf.keras.layers.Dense(10, activation=tf.nn.relu, input_shape=(4,)),\n",
        "    tf.keras.layers.Dense(10, activation=tf.nn.relu),\n",
        "    tf.keras.layers.Dense(3)\n",
        "])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "ok6y_qZ3LwbT",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        },
        "outputId": "71183987-f079-4bd6-8e08-798d5f81381e"
      },
      "cell_type": "code",
      "source": [
        "# Using the model\n",
        "predictions = model(features)\n",
        "predictions[:5]"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=207, shape=(5, 3), dtype=float32, numpy=\n",
              "array([[ 0.6022174 , -0.84608465, -0.16978204],\n",
              "       [ 0.59879875, -0.8756122 , -0.13861334],\n",
              "       [ 0.5448942 , -0.74825543, -0.41758808],\n",
              "       [ 0.70514214, -0.9941317 , -0.19577044],\n",
              "       [ 0.48032114, -1.0012796 , -0.62517655]], dtype=float32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "metadata": {
        "id": "bcLeyrsIMH9x",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "5309607f-b584-4ccc-ac98-43d72a627ebc"
      },
      "cell_type": "code",
      "source": [
        "# To convert these logits to a probability for each class, use the softmax function\n",
        "tf.nn.softmax(predictions[:5])\n",
        "\n",
        "# Taking the tf.argmax across classes gives us the predicted class index\n",
        "print(\"Prediction: {}\".format(tf.argmax(predictions, axis=1)))\n",
        "print(\"    Labels: {}\".format(labels))"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Prediction: [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]\n",
            "    Labels: [1 1 0 1 0 2 2 2 0 0 0 1 1 2 2 2 2 0 0 0 1 2 2 1 1 2 0 0 1 0 1 2]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "HScyK5J7Me94",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "e5d12a41-ff77-4feb-e433-818a09f2b9e4"
      },
      "cell_type": "code",
      "source": [
        "# Train the model\n",
        "# Define the loss and gradient function\n",
        "def loss(x, y, model):\n",
        "  y_ = model(x)\n",
        "  return tf.losses.sparse_softmax_cross_entropy(labels=y, logits=y_)\n",
        "\n",
        "l = loss(features, labels, model)\n",
        "print(\"Loss test: {}\".format(l))"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Loss test: 1.17226243019104\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "PNbXg75yNJdH",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Use the tf.GradientTape context to calculate the gradients used to optimize our model\n",
        "def grad(model, inputs, targets):\n",
        "  with tf.GradientTape() as tape:\n",
        "    loss_value = loss(inputs, targets, model)\n",
        "  return loss_value, tape.gradient(loss_value, model.trainable_variables)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "Xi9SOIloNkHm",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Create an optimizer\n",
        "optimizer = tf.train.AdamOptimizer(learning_rate=0.01)\n",
        "\n",
        "# setup the optimizer and the global_step counter\n",
        "global_step = tf.train.get_or_create_global_step()\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "_0SFC-wFO25n",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "d6b11c4a-50d5-48c3-82bb-c89ad4861602"
      },
      "cell_type": "code",
      "source": [
        "# calculate a single optimization step\n",
        "loss_value, grads = grad(model, features, labels)\n",
        "\n",
        "print(\"Step: {}, Initial Loss: {}\".format(global_step.numpy(),\n",
        "                                          loss_value.numpy()))\n",
        "\n",
        "optimizer.apply_gradients(zip(grads, model.variables), global_step)\n",
        "print(\"Step: {},         Loss: {}\".format(global_step.numpy(),\n",
        "                                          loss(features, labels, model).numpy()))"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Step: 0, Initial Loss: 1.17226243019104\n",
            "Step: 1,         Loss: 1.0696243047714233\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "AIgulGRUhpto",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 102
        },
        "outputId": "2fac65e9-d4fa-4f6a-a845-303fcdd851c4"
      },
      "cell_type": "code",
      "source": [
        "# Training loop\n",
        "## Note: Rerunning this cell uses the same model variables\n",
        "\n",
        "# keep results for plotting\n",
        "train_loss_results = []\n",
        "train_accuracy_results = []\n",
        "\n",
        "num_epochs = 201\n",
        "\n",
        "for epoch in range(num_epochs):\n",
        "  epoch_loss_avg = tfe.metrics.Mean()\n",
        "  epoch_accuracy = tfe.metrics.Accuracy()\n",
        "\n",
        "  # Training loop - using batches of 32\n",
        "  for x, y in train_dataset:\n",
        "    # Optimize the model\n",
        "    loss_value, grads = grad(model, x, y)\n",
        "    optimizer.apply_gradients(zip(grads, model.variables),\n",
        "                              global_step)\n",
        "\n",
        "    # Track progress\n",
        "    epoch_loss_avg(loss_value)  # add current batch loss\n",
        "    # compare predicted label to actual label\n",
        "    epoch_accuracy(tf.argmax(model(x), axis=1, output_type=tf.int32), y)\n",
        "\n",
        "  # end epoch\n",
        "  train_loss_results.append(epoch_loss_avg.result())\n",
        "  train_accuracy_results.append(epoch_accuracy.result())\n",
        "  \n",
        "  if epoch % 50 == 0:\n",
        "    print(\"Epoch {:03d}: Loss: {:.3f}, Accuracy: {:.3%}\".format(epoch,\n",
        "                                                                epoch_loss_avg.result(),\n",
        "                                                                epoch_accuracy.result()))"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 000: Loss: 0.966, Accuracy: 68.333%\n",
            "Epoch 050: Loss: 0.083, Accuracy: 98.333%\n",
            "Epoch 100: Loss: 0.059, Accuracy: 98.333%\n",
            "Epoch 150: Loss: 0.069, Accuracy: 97.500%\n",
            "Epoch 200: Loss: 0.045, Accuracy: 99.167%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "agjvNd2iUGFn",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 558
        },
        "outputId": "2fca6455-745d-4253-f452-1d9629b3ef81"
      },
      "cell_type": "code",
      "source": [
        "# Visualize the loss function over time\n",
        "fig, axes = plt.subplots(2, sharex=True, figsize=(12, 8))\n",
        "fig.suptitle('Training Metrics')\n",
        "\n",
        "axes[0].set_ylabel(\"Loss\", fontsize=14)\n",
        "axes[0].plot(train_loss_results)\n",
        "\n",
        "axes[1].set_ylabel(\"Accuracy\", fontsize=14)\n",
        "axes[1].set_xlabel(\"Epoch\", fontsize=14)\n",
        "axes[1].plot(train_accuracy_results);"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtgAAAIdCAYAAAAH9goCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCrJ/OAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7fe73fd18ef0>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "Ps3_9dJ3Lodk",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Setup the test dataset\n",
        "test_url = \"http://download.tensorflow.org/data/iris_test.csv\"\n",
        "\n",
        "test_fp = tf.keras.utils.get_file(fname=os.path.basename(test_url),\n",
        "                                  origin=test_url)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "SRMWCu30bnxH",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "test_dataset = tf.contrib.data.make_csv_dataset(\n",
        "    test_fp,\n",
        "    batch_size, \n",
        "    column_names=column_names,\n",
        "    label_name='species',\n",
        "    num_epochs=1,\n",
        "    shuffle=False)\n",
        "\n",
        "test_dataset = test_dataset.map(pack_feature_vector)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "Tw03-MK1cYId",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "29f1a785-2bd3-4973-b664-157eed5c7d61"
      },
      "cell_type": "code",
      "source": [
        "# Evaluate the model on test dataset\n",
        "test_accuracy = tfe.metrics.Accuracy()\n",
        "\n",
        "for (x, y) in test_dataset:\n",
        "  logits = model(x)\n",
        "  prediction = tf.argmax(logits, axis=1, output_type=tf.int32)\n",
        "  test_accuracy(prediction, y)\n",
        "\n",
        "print(\"Test set accuracy: {:.3%}\".format(test_accuracy.result()))"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Test set accuracy: 96.667%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "C2ZbDfOpRKhW",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 544
        },
        "outputId": "c2f0e573-28cd-416c-b937-f1fc64d1d855"
      },
      "cell_type": "code",
      "source": [
        "# check last batch\n",
        "tf.stack([y,prediction],axis=1)"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tf.Tensor: id=128146, shape=(30, 2), dtype=int32, numpy=\n",
              "array([[1, 1],\n",
              "       [2, 2],\n",
              "       [0, 0],\n",
              "       [1, 1],\n",
              "       [1, 1],\n",
              "       [1, 1],\n",
              "       [0, 0],\n",
              "       [2, 1],\n",
              "       [1, 1],\n",
              "       [2, 2],\n",
              "       [2, 2],\n",
              "       [0, 0],\n",
              "       [2, 2],\n",
              "       [1, 1],\n",
              "       [1, 1],\n",
              "       [0, 0],\n",
              "       [1, 1],\n",
              "       [0, 0],\n",
              "       [0, 0],\n",
              "       [2, 2],\n",
              "       [0, 0],\n",
              "       [1, 1],\n",
              "       [2, 2],\n",
              "       [1, 1],\n",
              "       [1, 1],\n",
              "       [1, 1],\n",
              "       [0, 0],\n",
              "       [1, 1],\n",
              "       [2, 2],\n",
              "       [1, 1]], dtype=int32)>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "metadata": {
        "id": "kesTS5Lzv-M2",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "outputId": "7fd285fc-eff0-4e9c-9938-70e6d7afbd03"
      },
      "cell_type": "code",
      "source": [
        "# Use the trained model to make predictions\n",
        "predict_dataset = tf.convert_to_tensor([\n",
        "    [5.1, 3.3, 1.7, 0.5,],\n",
        "    [5.9, 3.0, 4.2, 1.5,],\n",
        "    [6.9, 3.1, 5.4, 2.1]\n",
        "])\n",
        "\n",
        "predictions = model(predict_dataset)\n",
        "\n",
        "for i, logits in enumerate(predictions):\n",
        "  class_idx = tf.argmax(logits).numpy()\n",
        "  p = tf.nn.softmax(logits)[class_idx]\n",
        "  name = class_name[class_idx]\n",
        "  print(\"Example {} prediction: {} ({:4.1f}%)\".format(i, name, 100*p))"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Example 0 prediction: Iris setosa (99.9%)\n",
            "Example 1 prediction: Iris versicolor (99.9%)\n",
            "Example 2 prediction: Iris virginica (98.6%)\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}