crop_type_mapping/notebooks/LightGBM.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "kernel_info": {
      "name": "query-planet37"
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "nteract": {
      "version": "0.15.0"
    },
    "colab": {
      "name": "LightGBM.ipynb",
      "provenance": [],
      "collapsed_sections": []
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "6JYSIu-905lM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "!pip install -q rasterio rasterstats geopandas treeinterpreter lightgbm"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "ZmQRJPcH0oNL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import geopandas as gpd\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "\n",
        "import lightgbm as lgb\n",
        "\n",
        "import rasterio\n",
        "import rasterstats\n",
        "from sklearn.metrics import confusion_matrix\n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "from os import path as op\n",
        "import pickle"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uw1MvNkL2BE5",
        "colab_type": "code",
        "outputId": "cfbbc90e-faf2-49c0-ffe9-84548d865d46",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive', force_remount=True)\n",
        "root_dir = \"/content/drive/My Drive\""
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cYYI1YfE0oNR",
        "colab_type": "text"
      },
      "source": [
        "# Random Forest Model for Crop Type and Land Classification\n",
        "\n",
        "Using data created by SERVIR East Africa, RCMRD, and FEWSNET, we demonstrate how to train a LightGBM classifier over Trans Nzoia county, Kenya."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "__o2qQdE0oNR",
        "colab_type": "code",
        "outputId": "8a034dec-c3b8-40a2-fca9-caaf425acdad",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        }
      },
      "source": [
        "# read in training data\n",
        "training_vectors = gpd.read_file(op.join(root_dir, 'servir_croptype_workshop_data/training_data.geojson'))\n",
        "training_vectors.head()"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>name</th>\n",
              "      <th>description</th>\n",
              "      <th>geometry</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Shadow</td>\n",
              "      <td>None</td>\n",
              "      <td>MULTIPOLYGON (((34.83383 1.18204, 34.83577 1.1...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Forestland</td>\n",
              "      <td>None</td>\n",
              "      <td>MULTIPOLYGON (((35.30961 1.01328, 35.30964 1.0...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Maize</td>\n",
              "      <td>early reproductive</td>\n",
              "      <td>MULTIPOLYGON (((34.90904 1.09515, 34.90907 1.0...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Sugarcane</td>\n",
              "      <td>no change..maize farm on the right and far lef...</td>\n",
              "      <td>MULTIPOLYGON (((34.90750 1.08934, 34.90753 1.0...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Maize</td>\n",
              "      <td>reproductive good crop</td>\n",
              "      <td>MULTIPOLYGON (((34.87144 0.82953, 34.87147 0.8...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "         name  ...                                           geometry\n",
              "0      Shadow  ...  MULTIPOLYGON (((34.83383 1.18204, 34.83577 1.1...\n",
              "1  Forestland  ...  MULTIPOLYGON (((35.30961 1.01328, 35.30964 1.0...\n",
              "2       Maize  ...  MULTIPOLYGON (((34.90904 1.09515, 34.90907 1.0...\n",
              "3   Sugarcane  ...  MULTIPOLYGON (((34.90750 1.08934, 34.90753 1.0...\n",
              "4       Maize  ...  MULTIPOLYGON (((34.87144 0.82953, 34.87147 0.8...\n",
              "\n",
              "[5 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "LFMpu1PC0oNV",
        "colab_type": "code",
        "outputId": "dc6b6d9f-6bec-4275-9188-7e27f19985f6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "# find all unique values of training data names to use as classes\n",
        "classes = np.unique(training_vectors.name)\n",
        "# classes = np.array(sorted(training_vectors.name.unique()))\n",
        "classes"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array(['Built', 'Cloud', 'Fallow', 'Forestland', 'Grassland', 'Maize',\n",
              "       'Shadow', 'Sugarcane', 'Sunflower', 'Waterbody'], dtype=object)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "2FRc2gpg0oNX",
        "colab_type": "code",
        "outputId": "43bf9096-5a6e-4123-c07b-a5852f68fb50",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        }
      },
      "source": [
        "# create a dictionary to convert class names into integers for modeling\n",
        "class_dict = dict(zip(classes, range(len(classes))))\n",
        "class_dict                  "
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'Built': 0,\n",
              " 'Cloud': 1,\n",
              " 'Fallow': 2,\n",
              " 'Forestland': 3,\n",
              " 'Grassland': 4,\n",
              " 'Maize': 5,\n",
              " 'Shadow': 6,\n",
              " 'Sugarcane': 7,\n",
              " 'Sunflower': 8,\n",
              " 'Waterbody': 9}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "Tuz2ghqI0oNY",
        "colab_type": "code",
        "outputId": "1a7b2de9-2838-4618-8e5f-1503496573b5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "%%time\n",
        "# this larger cell reads data from a raster file for each training vector\n",
        "import rasterio\n",
        "from rasterio.features import rasterize\n",
        "from rasterstats.io import bounds_window\n",
        "# raster information\n",
        "raster_file = op.join(root_dir, 'servir_croptype_workshop_data/Trans_nzoia_2019_05-02.tif')\n",
        "bands = 6\n",
        "\n",
        "# a custom function for getting each value from the raster\n",
        "def all_values(x):\n",
        "    return x\n",
        "\n",
        "# this larger cell reads data from a raster file for each training vector\n",
        "X_raw = []\n",
        "y_raw = []\n",
        "with rasterio.open(raster_file, 'r') as src:\n",
        "    for (label, geom) in zip(training_vectors.name, training_vectors.geometry):\n",
        "        # read the raster data matching the geometry bounds\n",
        "        window = bounds_window(geom.bounds, src.transform)\n",
        "        # store our window information\n",
        "        window_affine = src.window_transform(window)\n",
        "        fsrc = src.read(window=window)\n",
        "        # rasterize the (non-buffered) geometry into the larger shape and affine\n",
        "        mask = rasterize(\n",
        "            [(geom, 1)],\n",
        "            out_shape=fsrc.shape[1:],\n",
        "            transform=window_affine,\n",
        "            fill=0,\n",
        "            dtype='uint8',\n",
        "            all_touched=True\n",
        "        ).astype(bool)\n",
        "        # for each label pixel (places where the mask is true)...\n",
        "        label_pixels = np.argwhere(mask)\n",
        "        for (row, col) in label_pixels:\n",
        "            # add a pixel of data to X\n",
        "            data = fsrc[:,row,col]\n",
        "            one_x = np.nan_to_num(data, nan=1e-3)\n",
        "            X_raw.append(one_x)\n",
        "            # add the label to y\n",
        "            y_raw.append(class_dict[label])"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 13.2 s, sys: 457 ms, total: 13.6 s\n",
            "Wall time: 20.4 s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "TyvZj0F_0oNa",
        "colab_type": "code",
        "outputId": "9626e7f6-0d3d-42c0-b976-98cf8972e141",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# convert the training data lists into the appropriate shape and format for scikit-learn\n",
        "X = np.array(X_raw)\n",
        "y = np.array(y_raw)\n",
        "(X.shape, y.shape)"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((160461, 6), (160461,))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "tJ4YWY5M0oNc",
        "colab_type": "code",
        "outputId": "fe72662c-b592-475e-9f8e-1f6644db4cd2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# (optional) add extra band indices\n",
        "\n",
        "# helper function for calculating ND*I indices (bands in the final dimension)\n",
        "def band_index(arr, a, b):\n",
        "    return np.expand_dims((arr[..., a] - arr[..., b]) / (arr[..., a] + arr[..., b]), axis=1)\n",
        "\n",
        "ndvi = band_index(X, 3, 2)\n",
        "ndwi = band_index(X, 1, 3)\n",
        "\n",
        "X = np.concatenate([X, ndvi, ndwi], axis=1)\n",
        "X.shape"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(160461, 8)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "6agtXCDh0oNe",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# split the data into test and train sets\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "9QetQBP40oNh",
        "colab_type": "code",
        "outputId": "25b263d6-8360-451f-e1e1-c22fc57bc019",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        }
      },
      "source": [
        "# calculate class weights to allow for training on inbalanced training samples\n",
        "labels, counts = np.unique(y_train, return_counts=True)\n",
        "class_weight_dict = dict(zip(labels, 1 / counts))\n",
        "class_weight_dict"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{0: 0.00046882325363338024,\n",
              " 1: 0.001597444089456869,\n",
              " 2: 0.0004928536224741252,\n",
              " 3: 1.970093973482535e-05,\n",
              " 4: 0.000819000819000819,\n",
              " 5: 1.5704257424187697e-05,\n",
              " 6: 0.0002473410833539451,\n",
              " 7: 0.0002824858757062147,\n",
              " 8: 0.05263157894736842,\n",
              " 9: 0.003115264797507788}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "JSuV7EVB0oNk",
        "colab_type": "code",
        "outputId": "68924c01-e1ce-4943-eb9f-ce90288b201d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "%%time\n",
        "# initialize a lightgbm\n",
        "\n",
        "\n",
        "lgbm = lgb.LGBMClassifier(\n",
        "       objective='multiclass',\n",
        "       class_weight = class_weight_dict,\n",
        "       num_class = len(class_dict),\n",
        "       metric = 'multi_logloss')"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 79 µs, sys: 5 µs, total: 84 µs\n",
            "Wall time: 90.4 µs\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "LXNiSfos0oNm",
        "colab_type": "code",
        "outputId": "879b428c-3975-4d37-eea6-c08c45b1a01a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 221
        }
      },
      "source": [
        "# fit the model to the data (training)\n",
        "lgbm.fit(X_train, y_train)"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "LGBMClassifier(boosting_type='gbdt',\n",
              "               class_weight={0: 0.00046882325363338024, 1: 0.001597444089456869,\n",
              "                             2: 0.0004928536224741252, 3: 1.970093973482535e-05,\n",
              "                             4: 0.000819000819000819, 5: 1.5704257424187697e-05,\n",
              "                             6: 0.0002473410833539451, 7: 0.0002824858757062147,\n",
              "                             8: 0.05263157894736842, 9: 0.003115264797507788},\n",
              "               colsample_bytree=1.0, importance_type='split', learning_rate=0.1,\n",
              "               max_depth=-1, metric='multi_logloss', min_child_samples=20,\n",
              "               min_child_weight=0.001, min_split_gain=0.0, n_estimators=100,\n",
              "               n_jobs=-1, num_class=10, num_leaves=31, objective='multiclass',\n",
              "               random_state=None, reg_alpha=0.0, reg_lambda=0.0, silent=True,\n",
              "               subsample=1.0, subsample_for_bin=200000, subsample_freq=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "25pMSxGp0oNo",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# predict on X_test to evaluate the model\n",
        "preds = lgbm.predict(X_test)\n",
        "cm = confusion_matrix(y_test, preds, labels=labels)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EhFM39HK-5NC",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model_name = 'light_gbm.sav'\n",
        "pickle.dump(lgbm, open(op.join(root_dir, model_name), 'wb'))\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "COjyzKe00oNq",
        "colab_type": "code",
        "outputId": "305ff914-83c5-4043-89f5-0a26d2ffc84f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 729
        }
      },
      "source": [
        "# plot the confusion matrix\n",
        "%matplotlib inline\n",
        "cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
        "fig, ax = plt.subplots(figsize=(10, 10))\n",
        "im = ax.imshow(cm, interpolation='nearest', cmap=plt.cm.Blues)\n",
        "ax.figure.colorbar(im, ax=ax)\n",
        "# We want to show all ticks...\n",
        "ax.set(xticks=np.arange(cm.shape[1]),\n",
        "       yticks=np.arange(cm.shape[0]),\n",
        "       # ... and label them with the respective list entries\n",
        "       xticklabels=classes, yticklabels=classes,\n",
        "       title='Normalized Confusion Matrix',\n",
        "       ylabel='True label',\n",
        "       xlabel='Predicted label')\n",
        "\n",
        "# Rotate the tick labels and set their alignment.\n",
        "plt.setp(ax.get_xticklabels(), rotation=45, ha=\"right\",\n",
        "         rotation_mode=\"anchor\")\n",
        "\n",
        "# Loop over data dimensions and create text annotations.\n",
        "fmt = '.2f'\n",
        "thresh = cm.max() / 2.\n",
        "for i in range(cm.shape[0]):\n",
        "    for j in range(cm.shape[1]):\n",
        "        ax.text(j, i, format(cm[i, j], fmt),\n",
        "                ha=\"center\", va=\"center\",\n",
        "                color=\"white\" if cm[i, j] > thresh else \"black\")\n",
        "fig.tight_layout()"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAqQAAALICAYAAACpcOKIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nlFKj/j/C8AIPDiwfgwAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 720x720 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jF5FtgwM0oN0",
        "colab_type": "text"
      },
      "source": [
        "## Generate predictions over the full image"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FKCEa-hXqgmN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# if want to use the pretrained model for new imagery\n",
        "\n",
        "# helper function for calculating ND*I indices (bands in the final dimension)\n",
        "# match the pretrained model weight with the saved model above\n",
        "model_name = 'light_gbm.sav'\n",
        "\n",
        "def band_index(arr, a, b):\n",
        "    return np.expand_dims((arr[..., a] - arr[..., b]) / (arr[..., a] + arr[..., b]), axis=1)\n",
        "  \n",
        "lgbm = pickle.load(open(op.join(root_dir, model_name), 'rb'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kbVo745N-S9G",
        "colab_type": "code",
        "outputId": "c590b681-fd40-4bb9-fce9-6daefa096c10",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 221
        }
      },
      "source": [
        "lgbm"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "LGBMClassifier(boosting_type='gbdt',\n",
              "               class_weight={0: 0.00046882325363338024, 1: 0.001597444089456869,\n",
              "                             2: 0.0004928536224741252, 3: 1.970093973482535e-05,\n",
              "                             4: 0.000819000819000819, 5: 1.5704257424187697e-05,\n",
              "                             6: 0.0002473410833539451, 7: 0.0002824858757062147,\n",
              "                             8: 0.05263157894736842, 9: 0.003115264797507788},\n",
              "               colsample_bytree=1.0, importance_type='split', learning_rate=0.1,\n",
              "               max_depth=-1, metric='multi_logloss', min_child_samples=20,\n",
              "               min_child_weight=0.001, min_split_gain=0.0, n_estimators=100,\n",
              "               n_jobs=-1, num_class=10, num_leaves=31, objective='multiclass',\n",
              "               random_state=None, reg_alpha=0.0, reg_lambda=0.0, silent=True,\n",
              "               subsample=1.0, subsample_for_bin=200000, subsample_freq=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QEA_rsWuCmi_",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# src.close()\n",
        "# dst.close()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "inputHidden": false,
        "outputHidden": false,
        "id": "vmD5HXr50oN8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# if want to use the pretrained model for new imagery\n",
        "# The pretrained model is called \"random_forest.sav\"\n",
        "# helper function for calculating ND*I indices (bands in the final dimension)\n",
        "\n",
        "# open connections to our input and output images\n",
        "\n",
        "# new_image = op.join(root_dir, 'Trans_nzoia_2019_10-04.tif')\n",
        "new_image = raster_file\n",
        "output_image = op.join(root_dir, \"lgbm_classification.tif\")\n",
        "src = rasterio.open(new_image, 'r')\n",
        "profile = src.profile\n",
        "profile.update(\n",
        "    dtype=rasterio.uint8,\n",
        "    count=1,\n",
        ")\n",
        "dst = rasterio.open(output_image, 'w', **profile)\n",
        "\n",
        "# perform prediction on each small image patch to minimize required memory\n",
        "patch_size = 500\n",
        "\n",
        "for i in range((src.shape[0] // patch_size) + 1):\n",
        "    for j in range((src.shape[1] // patch_size) + 1):\n",
        "        # define the pixels to read (and write)\n",
        "        window = rasterio.windows.Window(\n",
        "            j * patch_size,\n",
        "            i * patch_size,\n",
        "            # don't read past the image bounds\n",
        "            min(patch_size, src.shape[1] - j * patch_size),\n",
        "            min(patch_size, src.shape[0] - i * patch_size)\n",
        "        )\n",
        "        \n",
        "        data = src.read(window=window)\n",
        "        # read the image into the proper format, adding indices if necessary\n",
        "        img_swp = np.moveaxis(data, 0, 2)\n",
        "        img_flat = img_swp.reshape(-1, img_swp.shape[-1])\n",
        "\n",
        "        img_ndvi = band_index(img_flat, 3, 2)\n",
        "        img_ndwi = band_index(img_flat, 1, 3)\n",
        "\n",
        "        img_w_ind = np.concatenate([img_flat, img_ndvi, img_ndwi], axis=1)\n",
        "\n",
        "        # remove no data values, store the indices for later use\n",
        "        # a later cell makes the assumption that all bands have identical no-data value arrangements\n",
        "        m = np.ma.masked_invalid(img_w_ind)\n",
        "        to_predict = img_w_ind[~m.mask].reshape(-1, img_w_ind.shape[-1])\n",
        "        \n",
        "        # predict\n",
        "        if not len(to_predict):\n",
        "            continue\n",
        "        img_preds = lgbm.predict(to_predict)\n",
        "        \n",
        "        # add the prediction back to the valid pixels (using only the first band of the mask to decide on validity)\n",
        "        # resize to the original image dimensions\n",
        "        output = np.zeros(img_flat.shape[0])\n",
        "        output[~m.mask[:,0]] = img_preds.flatten()\n",
        "        output = output.reshape(*img_swp.shape[:-1])\n",
        "        \n",
        "        # create our final mask\n",
        "        mask = (~m.mask[:,0]).reshape(*img_swp.shape[:-1])\n",
        "\n",
        "        # write to the final file\n",
        "        dst.write(output.astype(rasterio.uint8), 1, window=window)\n",
        "        dst.write_mask(mask, window=window)\n",
        "        # write to the final file\n",
        "        dst.write(output.astype(rasterio.uint8), 1, window=window)\n",
        "        dst.write_mask(mask, window=window)\n",
        "\n",
        "src.close()\n",
        "dst.close()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rTORxEa038_C",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}