CV512077-NN_integer_detection.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "CV512077.ipynb",
      "provenance": [],
      "authorship_tag": "ABX9TyO8CEEd79HnnYXnEi/p9NtM",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/bmreiniger/datascience.stackexchange/blob/master/CV512077-NN_integer_detection.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Slrrrg0xe3IO"
      },
      "source": [
        "from sklearn.neural_network import MLPClassifier\n",
        "from sklearn.pipeline import Pipeline\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.model_selection import train_test_split\n",
        "import numpy as np\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FpYlkIM6gHsu"
      },
      "source": [
        "ints = np.hstack( (np.arange(6), np.arange(7,10)) )  # skip 6\n",
        "ints = np.hstack( (ints,) * 10)  # for balance...skl.mlp doesn't have class_weights\n",
        "other = np.random.random(100)*10\n",
        "X = np.hstack((ints, other))[:, np.newaxis]\n",
        "y = np.array([1]*len(ints) + [0]*len(other))\n",
        "w = np.array([len(other)/len(ints)]*len(ints) + [1]*len(other))\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cwFnUfLhlWx2",
        "outputId": "f5f542bd-d202-490a-a15b-f69654fd2472"
      },
      "source": [
        "ints"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([0, 1, 2, 3, 4, 5, 7, 8, 9, 0, 1, 2, 3, 4, 5, 7, 8, 9, 0, 1, 2, 3,\n",
              "       4, 5, 7, 8, 9, 0, 1, 2, 3, 4, 5, 7, 8, 9, 0, 1, 2, 3, 4, 5, 7, 8,\n",
              "       9, 0, 1, 2, 3, 4, 5, 7, 8, 9, 0, 1, 2, 3, 4, 5, 7, 8, 9, 0, 1, 2,\n",
              "       3, 4, 5, 7, 8, 9, 0, 1, 2, 3, 4, 5, 7, 8, 9, 0, 1, 2, 3, 4, 5, 7,\n",
              "       8, 9])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 34
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Llwmegp3gKSN"
      },
      "source": [
        "X_train, X_test, y_train, y_test, w_train, w_test = train_test_split(X, y, w, stratify=y, test_size=0.1)\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LMiEmFDMhJ7D",
        "outputId": "f61ba631-648e-4d2e-cb30-56abb9d07e62"
      },
      "source": [
        "mlp = Pipeline([('scale', StandardScaler()), ('mlp', MLPClassifier((40, 15), solver='lbfgs', max_iter=10000, tol=0.00001, learning_rate_init=0.0001))])\n",
        "mlp.fit(X_train, y_train)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Pipeline(memory=None,\n",
              "         steps=[('scale',\n",
              "                 StandardScaler(copy=True, with_mean=True, with_std=True)),\n",
              "                ('mlp',\n",
              "                 MLPClassifier(activation='relu', alpha=0.0001,\n",
              "                               batch_size='auto', beta_1=0.9, beta_2=0.999,\n",
              "                               early_stopping=False, epsilon=1e-08,\n",
              "                               hidden_layer_sizes=(40, 15),\n",
              "                               learning_rate='constant',\n",
              "                               learning_rate_init=0.0001, max_fun=15000,\n",
              "                               max_iter=10000, momentum=0.9,\n",
              "                               n_iter_no_change=10, nesterovs_momentum=True,\n",
              "                               power_t=0.5, random_state=None, shuffle=True,\n",
              "                               solver='lbfgs', tol=1e-05,\n",
              "                               validation_fraction=0.1, verbose=False,\n",
              "                               warm_start=False))],\n",
              "         verbose=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 59
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JdIsBHtPkgV9",
        "outputId": "b78d7a4e-aaab-4239-8f41-c0ffdbf11397"
      },
      "source": [
        "mlp.score(X_train, y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.7953216374269005"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 60
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NwdPC4hthRg6",
        "outputId": "79e8f35e-80a0-4066-ebea-db12292a5d4b"
      },
      "source": [
        "mlp.score(X_test, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.6842105263157895"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 61
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Xfs3Fwd2hTPr",
        "outputId": "58c26665-4045-40d7-8dc9-f7864f549ea1"
      },
      "source": [
        "y_test"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([1, 0, 0, 0, 0, 0])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 69
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "efxnt5gAhVOB",
        "outputId": "f289ea05-4d7f-43d1-8c9c-453fa065502c"
      },
      "source": [
        "mlp.predict(X_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([1, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 1, 0, 0, 1, 0])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "id": "iGes1f0vhbA1",
        "outputId": "9c12cd89-8ba5-4c8d-b271-6788adb73365"
      },
      "source": [
        "from matplotlib import pyplot as plt\n",
        "xx = np.linspace(-3, 13, num=1000)\n",
        "yy = mlp.predict_proba(xx[:, np.newaxis])[:, 1]\n",
        "plt.plot(xx, yy)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<matplotlib.lines.Line2D at 0x7f445fdbde10>]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 62
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bMgyZZs_kYHq"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}