tutorials/explainers/LIME/lime_timeseries_coffee.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img width=\"150\" alt=\"Logo_ER10\" src=\"https://user-images.githubusercontent.com/3244249/151994514-b584b984-a148-4ade-80ee-0f88b0aefa45.png\">\n",
    "\n",
    "### Model Interpretation using LIME for coffee dataset classifier\n",
    "This notebook demonstrates the use of DIANNA with the LIME timeseries method on the [Coffee dataset](https://www.timeseriesclassification.com/description.php?Dataset=Coffee).\n",
    "\n",
    "LIME (Local Interpretable Model-agnostic Explanations) is an explainable-AI method that aims to create an interpretable model that locally represents the classifier. For more details see the [LIME paper](https://arxiv.org/abs/1602.04938)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Colab Setup"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-31T09:48:42.651335Z",
     "start_time": "2023-05-31T09:48:42.646323Z"
    }
   },
   "outputs": [],
   "source": [
    "running_in_colab = 'google.colab' in str(get_ipython())\n",
    "if running_in_colab:\n",
    "    # install dianna\n",
    "    !python3 -m pip install dianna[notebooks]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Import libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-31T09:48:43.874520Z",
     "start_time": "2023-05-31T09:48:42.653341Z"
    }
   },
   "outputs": [],
   "source": [
    "import dianna\n",
    "import pandas as pd\n",
    "from sklearn.neighbors import KNeighborsClassifier as KNN\n",
    "from sklearn.metrics import accuracy_score as acc\n",
    "import numpy as np\n",
    "from dianna.utils.downloader import download"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1 - Loading the data\n",
    "Load coffee dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-31T09:48:43.909870Z",
     "start_time": "2023-05-31T09:48:43.878877Z"
    }
   },
   "outputs": [],
   "source": [
    "# Load coffee dataset\n",
    "coffee_train = pd.read_csv(download('coffee_train.csv', 'data'),\n",
    "                            sep=',', header=None).astype(float)\n",
    "coffee_train_y = coffee_train.loc[:, 0]\n",
    "coffee_train_x = coffee_train.loc[:, 1:]\n",
    "coffee_test = pd.read_csv(download('coffee_test.csv', 'data'),\n",
    "                           sep=',', header=None).astype(float)\n",
    "coffee_test_y = coffee_test.loc[:, 0]\n",
    "coffee_test_x = coffee_test.loc[:, 1:]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2 - Define a KNN model to classify two types of coffee\n",
    "Here we train a simple k-nearest neighbors (KNN) model for classification task."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-31T09:48:43.996106Z",
     "start_time": "2023-05-31T09:48:43.909870Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy KNN for coffee dataset: 0.964286\n"
     ]
    }
   ],
   "source": [
    "# Define and train model\n",
    "knn = KNN()\n",
    "knn.fit(coffee_train_x, coffee_train_y)\n",
    "print('Accuracy KNN for coffee dataset: %f' % (acc(coffee_test_y, knn.predict(coffee_test_x))))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "An instance is selected for explanation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-31T09:48:44.015023Z",
     "start_time": "2023-05-31T09:48:43.999815Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(286,)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Select instance for explanation\n",
    "idx = 5 # explained instance\n",
    "series = coffee_test_x.iloc[idx, :]\n",
    "num_features = len(series) # how many feature contained in explanation\n",
    "num_slices = len(series) # split time series\n",
    "\n",
    "class_names = ['Arabica', 'Robusta']\n",
    "labels = np.argsort(class_names)\n",
    "series.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that the LIME timeseries explainer in DIANNA can only receive input in the order of [timeseries, channels]. Given the way that the model was trained, we need to create an model runner for the explainer."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-31T09:48:44.069544Z",
     "start_time": "2023-05-31T09:48:44.013016Z"
    }
   },
   "outputs": [],
   "source": [
    "# the explainer must receive input in the order of [timeseries, channels]\n",
    "data_instance = np.expand_dims(series, axis=1)\n",
    "def run_model(data):\n",
    "    data = data[:,:,0]\n",
    "    return knn.predict_proba(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3 - Applying LIME with DIANNA for timeseries"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this section we apply LIME to this KNN model to calculate the relevance score and visualize the results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-31T09:48:46.604782Z",
     "start_time": "2023-05-31T09:48:44.043890Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Explaining: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 500/500 [00:00<00:00, 1386.36it/s]\n"
     ]
    }
   ],
   "source": [
    "exp = dianna.explain_timeseries(run_model, input_timeseries=data_instance, method='lime',\n",
    "                                labels=[0,1], class_names=class_names, num_features=num_features,\n",
    "                                num_samples=500, num_slices=num_slices, distance_method='cosine')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can visualize results using the built-in visualizer in `LIME` explainer. The elements contributing positively towards the classification decision are indicated in 'red' and those contributing negatively - in 'blue'."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-31T09:48:47.575864Z",
     "start_time": "2023-05-31T09:48:46.604500Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from dianna import visualization\n",
    "# Normalize the explanation scores for the purpose of visualization\n",
    "def normalize(data):\n",
    "    \"\"\"Squash all values into [-1,1] range.\"\"\"\n",
    "    zero_to_one = (data - np.min(data)) / (np.max(data) - np.min(data))\n",
    "    return 2*zero_to_one -1\n",
    "\n",
    "heatmap_channel = normalize(exp[0])\n",
    "segments = []\n",
    "for i in range(len(heatmap_channel) - 1):\n",
    "    segments.append({\n",
    "        'index': i,\n",
    "        'start': i - 0.5,\n",
    "        'stop': i + 0.5,\n",
    "        'weight': heatmap_channel[i]})\n",
    "fig, _ = visualization.plot_timeseries(range(len(heatmap_channel)), data_instance,\n",
    "                              segments, x_label=\"Time index\", y_label=\"Temperature\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 4- Experiment with ONNX\n",
    "DIANNA supports ONNX models. Here we demonstrate the use of LIME explainer for timeseries with a pre-trained ONNX model, which is a random forest classifier for the same [Coffee dataset](https://www.timeseriesclassification.com/description.php?Dataset=Coffee). <br>\n",
    "\n",
    "The model is trained by this [notebook](https://github.com/dianna-ai/dianna-exploration/blob/main/example_data/model_generation/coffee/generate_model.ipynb): <br>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-31T09:48:47.637503Z",
     "start_time": "2023-05-31T09:48:47.578872Z"
    }
   },
   "outputs": [],
   "source": [
    "from dianna.utils.onnx_runner import SimpleModelRunner\n",
    "\n",
    "model_path = download('coffee.onnx', 'model')\n",
    "loaded_model = SimpleModelRunner(model_path)\n",
    "predictions = loaded_model(data_instance.T.astype(np.float32))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A runner function is required since the model works with input in the order of [channels, timeseries], while LIME in DIANNA should receive input in the opposite order. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-31T09:48:47.638506Z",
     "start_time": "2023-05-31T09:48:47.593279Z"
    }
   },
   "outputs": [],
   "source": [
    "import onnxruntime as ort\n",
    "\n",
    "def run_model(data):\n",
    "    # model must receive input in the order of [channels, timeseries]\n",
    "    # get ONNX predictions\n",
    "    predictions = []\n",
    "    for i in range(data.shape[0]):\n",
    "        input_data = data[i].T\n",
    "        sess = ort.InferenceSession(model_path)\n",
    "        input_name = sess.get_inputs()[0].name\n",
    "        output_name = sess.get_outputs()[0].name\n",
    "\n",
    "        onnx_input = {input_name: input_data.astype(np.float32)}\n",
    "        pred_onnx = sess.run([output_name], onnx_input)[0]\n",
    "        predictions.append(pred_onnx)\n",
    "    \n",
    "    return np.asarray(predictions)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can call the explainer and visualize the results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-31T09:48:49.417332Z",
     "start_time": "2023-05-31T09:48:47.613369Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Explaining: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 500/500 [00:01<00:00, 345.87it/s]\n"
     ]
    }
   ],
   "source": [
    "exp = dianna.explain_timeseries(run_model, input_timeseries=data_instance, method='lime',\n",
    "                                labels=[0], class_names=class_names, num_features=num_features,\n",
    "                                num_samples=500, num_slices=num_slices, batch_size=1, distance_method='cosine')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-05-31T09:48:50.058148Z",
     "start_time": "2023-05-31T09:48:49.419338Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "heatmap_channel = normalize(exp[0])\n",
    "segments = []\n",
    "for i in range(len(heatmap_channel) - 1):\n",
    "    segments.append({\n",
    "        'index': i,\n",
    "        'start': i - 0.5,\n",
    "        'stop': i + 0.5,\n",
    "        'weight': heatmap_channel[i]})\n",
    "fig, _ = visualization.plot_timeseries(range(len(heatmap_channel)), data_instance,\n",
    "                              segments, x_label=\"Time index\", y_label=\"Temperature\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 5 - Conclusions\n",
    "The importance of each segmententation of timeseries to the classification result is shown in the image. This is achieved by applying LIME approach to timeseries.\n",
    "\n",
    "Both examples ...(To be completed.)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}