notebooks/wheelie_label_ml.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "wheelie_label_ml.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "KMqJPaVZ9k-x",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "UPLOAD_FOLDER = 'My Drive/wheelie_training/'\n",
        "FIG_FOLDER = 'ml/label/'\n",
        "MODEL_FOLDER = 'label/'\n",
        "\n",
        "# data and training parameters\n",
        "GET_ONE_I_PREPARED_EARLIER = True\n",
        "session_count = 2\n",
        "PREVIOUS_TIMESTEP = True\n",
        "MAJORITY_UNDERSAMPLING = True\n",
        "NORMALISE = False\n",
        "\n",
        "import matplotlib\n",
        "primary_cmap = matplotlib.cm.get_cmap('cool')\n",
        "secondary_cmap = matplotlib.cm.get_cmap('Wistia')\n",
        "tertiary_cmap = matplotlib.cm.get_cmap('summer')"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "guVHd4em9l3U",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "GDRIVE_ROOT = '/content/gdrive/'\n",
        "FS_ROOT = GDRIVE_ROOT + UPLOAD_FOLDER\n",
        "RAW_ROOT = FS_ROOT\n",
        "CSV_EXT = '*.csv'\n",
        "PROC_ROOT = RAW_ROOT + 'processed/'\n",
        "FIG_ROOT = RAW_ROOT + 'figures/'\n",
        "FIG_DIR = FIG_ROOT + FIG_FOLDER\n",
        "MODEL_ROOT = RAW_ROOT + 'models/'\n",
        "MODEL_DIR = MODEL_ROOT + MODEL_FOLDER"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZcBKDgYVFWQj",
        "colab_type": "text"
      },
      "source": [
        "## Work with files"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VaZjK4jGRhDT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount(GDRIVE_ROOT)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-EQa_RH_91IJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import datetime\n",
        "import math\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "from pathlib import Path\n",
        "import pytz\n",
        "\n",
        "!pip install python-rle\n",
        "import rle\n",
        "\n",
        "def get_df_raw(fname):\n",
        "  return pd.read_csv(RAW_ROOT + fname)\n",
        "\n",
        "proc_dfs = {}\n",
        "def get_df_proc(fname):\n",
        "  if not fname in proc_dfs.keys():\n",
        "    proc_dfs[fname] = pd.read_csv(PROC_ROOT + fname)\n",
        "  return proc_dfs[fname]\n",
        "\n",
        "def put_df(df, fname):\n",
        "  Path(MODEL_DIR).mkdir(parents=True, exist_ok=True)\n",
        "  df.to_csv(MODEL_DIR + fname)\n",
        "\n",
        "def get_fname(file):\n",
        "  return file.split('/')[-1]\n",
        "\n",
        "def put_fig(name):\n",
        "  Path(FIG_DIR).mkdir(parents=True, exist_ok=True)\n",
        "  plt.savefig(FIG_DIR + name + '.png')\n",
        "\n",
        "def fromtimestamp_mel(ts):\n",
        "  return datetime.datetime.fromtimestamp(ts, tz=pytz.timezone('Australia/Melbourne'))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1DGgTYLZBOCQ",
        "colab_type": "text"
      },
      "source": [
        "## Get data and work with time"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ijsWRP0B96E2",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "accum = get_df_proc('accum.csv')\n",
        "if not 'wnum' in accum.columns:\n",
        "  accum.index = pd.Series(range(len(accum.index)), name='wnum')\n",
        "accum['datetime'] = accum['timestamp'].apply(fromtimestamp_mel)"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uRgpwwAFUNFs",
        "colab_type": "text"
      },
      "source": [
        "## ML utilities"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2-DWXUasULp1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.metrics import confusion_matrix, precision_recall_fscore_support, accuracy_score\n",
        "\n",
        "def cm_img(y_true, y_pred, ax, classes=[1, 0], desc='Evaluation'):\n",
        "  cmf = np.flip(confusion_matrix(y_true, y_pred))\n",
        "  fscore = precision_recall_fscore_support(y_true, y_pred, labels=classes, average='binary')[2]\n",
        "  ascore = accuracy_score(y_true, y_pred)\n",
        "  im = ax.imshow(cmf, cmap=secondary_cmap)\n",
        "  ax.xaxis.tick_top()\n",
        "  ax.xaxis.set_label_position('top') \n",
        "  ax.set_xticks(classes)\n",
        "  ax.set_yticks(classes)\n",
        "  ax.set_xticklabels(reversed(classes))\n",
        "  ax.set_yticklabels(reversed(classes))\n",
        "  ax.set_xlabel('Pred')\n",
        "  ax.set_ylabel('True')\n",
        "  ax.set_title('{}\\nAccuracy {:.3f}, F-Score {:.3f}'.format(desc, ascore, fscore), y=-0.3)\n",
        "  for i in range(cmf.shape[0]):\n",
        "    for j in range(cmf.shape[1]):\n",
        "      text = ax.text(j, i, cmf[i, j], ha=\"center\", va=\"center\")"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-fm42bVyBUse",
        "colab_type": "text"
      },
      "source": [
        "# Label wheelies with ML"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8Is5dNtiGHMw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# inputs to the model\n",
        "features = ['accelerometerAccelerationX',\n",
        "            'accelerometerAccelerationY',\n",
        "            'accelerometerAccelerationZ']\n",
        "\n",
        "# label we will train to\n",
        "original_label = 'wheel_up'"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4a9ouoIoGe9a",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# get some sessions data\n",
        "sessions_df = pd.DataFrame()\n",
        "if GET_ONE_I_PREPARED_EARLIER:\n",
        "  sessions_df = pd.read_csv(MODEL_DIR + 'accel_label_{}_sessions.csv'.format(session_count))\n",
        "else:\n",
        "  fnames = [get_fname(f) for f in accum['source'].unique()]\n",
        "  sessions_df = pd.concat([get_df_proc(f) for f in fnames[-session_count:]])\n",
        "  cache_df = sessions_df[features + [original_label]]\n",
        "  put_df(cache_df, 'accel_label_{}_sessions.csv'.format(session_count))"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "B8OA0OOG0gvq",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "y = np.array(sessions_df[original_label])\n",
        "X = np.array(sessions_df[features]).reshape(-1, len(features))"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FwXz8o81G-q_",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 193
        },
        "outputId": "28e7bea8-9e70-41e2-a0e1-1d7dfc288686"
      },
      "source": [
        "# add prev timestep\n",
        "# note we could drop rows where this rolls over discontinuities between sessions\n",
        "if PREVIOUS_TIMESTEP:\n",
        "  X_prev = np.roll(X, 1, axis=0)\n",
        "  X = np.concatenate((X_prev, X), axis=1)\n",
        "print('input shape: {}'.format(X.shape))\n",
        "\n",
        "# majority class undersampling\n",
        "from collections import Counter\n",
        "from imblearn.under_sampling import RandomUnderSampler\n",
        "print('Raw class breakdown: {}'.format(Counter(y)))\n",
        "if MAJORITY_UNDERSAMPLING:\n",
        "  undersample = RandomUnderSampler(random_state=42)\n",
        "  X, y = undersample.fit_resample(X, y)\n",
        "  print('After undersampling: {}'.format(Counter(y)))"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "input shape: (37758, 6)\n",
            "Raw class breakdown: Counter({0: 33831, 1: 3927})\n",
            "After undersampling: Counter({0: 3927, 1: 3927})\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/sklearn/externals/six.py:31: FutureWarning: The module is deprecated in version 0.21 and will be removed in version 0.23 since we've dropped support for Python 2.7. Please rely on the official version of six (https://pypi.org/project/six/).\n",
            "  \"(https://pypi.org/project/six/).\", FutureWarning)\n",
            "/usr/local/lib/python3.6/dist-packages/sklearn/utils/deprecation.py:144: FutureWarning: The sklearn.neighbors.base module is  deprecated in version 0.22 and will be removed in version 0.24. The corresponding classes / functions should instead be imported from sklearn.neighbors. Anything that cannot be imported from sklearn.neighbors is now part of the private API.\n",
            "  warnings.warn(message, FutureWarning)\n",
            "/usr/local/lib/python3.6/dist-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function safe_indexing is deprecated; safe_indexing is deprecated in version 0.22 and will be removed in version 0.24.\n",
            "  warnings.warn(msg, category=FutureWarning)\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cIzJrWtH-g0i",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 540
        },
        "outputId": "34f53d85-b22f-4c95-85ca-d92c14ab9af1"
      },
      "source": [
        "subplot_cols = len(features)\n",
        "subplot_rows = X.shape[1] // subplot_cols\n",
        "fig, ax = plt.subplots(subplot_rows, subplot_cols)\n",
        "fig.set_figwidth(12)\n",
        "fig.set_figheight(4 * subplot_rows)\n",
        "fig.suptitle('Accelerometer input distributions by wheelie class')\n",
        "for i in range(X.shape[1]):\n",
        "  ax_id = ax[i % subplot_cols] if subplot_rows == 1 else ax[i // subplot_cols][i % subplot_cols]\n",
        "  y_1_id = np.argwhere(y > 0)\n",
        "  y_0_id = np.argwhere(y < 1)\n",
        "  ax_id.hist(X[y_1_id, i], bins=np.linspace(-2, 2, 50), alpha=0.7, color=tertiary_cmap(0.2))\n",
        "  ax_id.hist(X[y_0_id, i], bins=np.linspace(-2, 2, 50), alpha=0.7, color=tertiary_cmap(0.8))\n",
        "  ax_id.set_title('F{} @ t-{}'.format(i % subplot_cols, i // subplot_cols))\n",
        "  ax_id.legend(['1', '0'])\n",
        "plt.show()"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x576 with 6 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ifE00SmKe1Au",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 294
        },
        "outputId": "81bad9f7-6b75-470b-ba9b-c01f470d62b0"
      },
      "source": [
        "def visual_scale_y(y_val):\n",
        "  return y_val * 8 - 4;\n",
        "\n",
        "index_accel_y = 1;\n",
        "index_accel_z = 2;\n",
        "fig, ax = plt.subplots(1, 2)\n",
        "fig.set_figwidth(12)\n",
        "fig.suptitle('Comparison of Y & Z accelerations with wheelie class')\n",
        "ax[0].plot(visual_scale_y(y))\n",
        "ax[0].plot(X[:,index_accel_z])\n",
        "ax[0].plot(X[:,index_accel_y])\n",
        "ax[0].set_title('By Sample')\n",
        "ax[0].legend(['class', 'y', 'z'])\n",
        "sc = ax[1].scatter(X[:,index_accel_y], X[:,index_accel_z], c=y, alpha=0.2)\n",
        "ax[1].set_title('y vs z, color by class')\n",
        "plt.colorbar(sc)\n",
        "plt.show()"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x288 with 3 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LFM7zH1tCoCD",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.20, random_state=42)"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aTqFe8RQ-eVa",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 139
        },
        "outputId": "ab94518a-622d-4ebd-f8c8-f4eeaffe68af"
      },
      "source": [
        "# normalise\n",
        "from scipy.stats import norm\n",
        "norm_params = np.array([norm.fit(X_train[:,i]) for i in range(X.shape[1])])\n",
        "if NORMALISE:\n",
        "  X_train = (X_train - norm_params[:,0]) / norm_params[:,1]\n",
        "  X_test = (X_test - norm_params[:,0]) / norm_params[:,1]\n",
        "print('Performed normalisation: {}'.format(NORMALISE))\n",
        "print(norm_params)"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Performed normalisation: False\n",
            "[[ 0.04184072  0.23712943]\n",
            " [-0.75674233  0.31702173]\n",
            " [-0.53290722  0.42451764]\n",
            " [ 0.04262741  0.24618426]\n",
            " [-0.75393035  0.30171027]\n",
            " [-0.53425504  0.44859057]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PqR4GiNbltEP",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "param_desc = 't-1: {}, under: {}, norm: {}'.format(PREVIOUS_TIMESTEP, MAJORITY_UNDERSAMPLING, NORMALISE)"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ocfOfQwEDDtx",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 340
        },
        "outputId": "467a860b-c729-4b0b-ed03-85cd3a315abc"
      },
      "source": [
        "from sklearn.linear_model import LogisticRegression\n",
        "\n",
        "label_model = LogisticRegression(solver='lbfgs')\n",
        "label_model.fit(X_train, y_train)\n",
        "\n",
        "y_pred = label_model.predict(X_test)\n",
        "cm_img(y_test, y_pred, plt.gca(), desc='LR wheelie from accel\\n{}'.format(param_desc))\n",
        "put_fig('log_regr_conf')\n",
        "plt.show()"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "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": "ycphSviscXUg",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 278
        },
        "outputId": "dd76335e-eccc-411f-8c18-40d244360115"
      },
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow.keras import layers\n",
        "from tensorflow.keras import losses\n",
        "\n",
        "x_tf = tf.keras.Input(shape=(6,))\n",
        "y_tf_predicted = x_tf\n",
        "y_tf_predicted = layers.Dense(12, activation='relu')(y_tf_predicted)\n",
        "y_tf_predicted = layers.Dense(1, activation='sigmoid')(y_tf_predicted)\n",
        "nn_model = tf.keras.Model(inputs=x_tf, outputs=y_tf_predicted)\n",
        "print(nn_model.summary())"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"functional_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_1 (InputLayer)         [(None, 6)]               0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 12)                84        \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 1)                 13        \n",
            "=================================================================\n",
            "Total params: 97\n",
            "Trainable params: 97\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GZjboVPVcoSQ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        },
        "outputId": "feed0702-427d-4af1-fd85-b98a36d1cf83"
      },
      "source": [
        "nn_model.compile(optimizer=tf.optimizers.SGD(learning_rate=1.0),\n",
        "              loss=losses.binary_crossentropy)\n",
        "history = nn_model.fit(x=X_train, y=y_train, epochs=100, verbose=0)\n",
        "plt.plot(history.history['loss'])\n",
        "plt.title(\"loss over training\")\n",
        "plt.show()"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "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": "GSyrAOWtd4rP",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 340
        },
        "outputId": "06bf2bc5-f33b-4992-badd-efbeabd18b47"
      },
      "source": [
        "y_pred_tf = nn_model.predict(x=X_test)\n",
        "\n",
        "def threshold(pred):\n",
        "  return pred > 0.5\n",
        "\n",
        "cm_img(y_test, [threshold(y) for y in y_pred_tf], plt.gca(), desc='NN wheelie from accel\\n{}'.format(param_desc))\n",
        "put_fig('nn_conf')\n",
        "plt.show()"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "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": "rDzckcqLmt33",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import h5py\n",
        "nn_model.save(MODEL_DIR + 'keras_model_2.h5')"
      ],
      "execution_count": 20,
      "outputs": []
    }
  ]
}