5-Evaluation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "dac66dbc-7ba9-4115-98f7-f2470fb4b844",
   "metadata": {},
   "source": [
    "# Evaluation"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "da94be4a-3bc6-4ee8-aee8-76ff5de0c030",
   "metadata": {},
   "source": [
    "In this hackathon, you will use this notebook to evaluate the performance of your machine learning pipeline. The notebook is designed to assess the entire pipeline, from data preprocessing to making predictions, ensuring that your solution is practical, efficient, and accurate.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a5505d86-e261-4e07-98be-d9246016d41e",
   "metadata": {},
   "source": [
    "**Important Note:**\n",
    "- The evaluation runs on a `single` CPU core to ensure fair resource usage comparisons.\n",
    "- It assesses the pipeline’s performance as a whole, not just the final model’s predictions."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ec781abf-7ce3-4740-8a88-35e9aae2e2a5",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2cecd5e0-992a-442a-8147-b0c3231949be",
   "metadata": {},
   "source": [
    "### 1. What the Evaluation Measures"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "984aeb69-5f5b-4aa2-ae4a-222149f9a993",
   "metadata": {},
   "source": [
    "The notebook evaluates your pipeline on several metrics:"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dcf5fef0-5fa3-48f2-ae5d-ca7554e21c4f",
   "metadata": {},
   "source": [
    "#### **1.\tAccuracy and Model Performance:**"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b893966e-d7c6-477d-a519-3332eab7ac82",
   "metadata": {},
   "source": [
    "- **Accuracy**: How many predictions were correct.\n",
    "- **F1 Score**: A balanced measure of precision and recall, especially for imbalanced datasets.\n",
    "- **Confusion Matrix**: A detailed breakdown of your model’s predictions across the following classes:\n",
    "    - `Priority`: The most important class, representing clear and usable images. Misclassifying these has the greatest negative impact.\n",
    "\t- `Noisy` & `Blurry`: Impure images that could potentially be recovered through preprocessing. Errors here are less critical but still significant.\n",
    "\t- `Corrupt` & `Missing_Data`: The least important classes, representing images with severe issues or no usable data. Misclassifications here have the smallest impact."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eeb7d683-ecd4-4f7d-a1b2-fc2f36783b80",
   "metadata": {},
   "source": [
    "#### **2. Evaluation Time:**"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f3d330de-8029-4463-8eeb-cfe24768287d",
   "metadata": {},
   "source": [
    "- **What we measure**: How long your pipeline takes to preprocess the data and make predictions.\n",
    "- **Single CPU Core**: The evaluation runs on a `single` CPU core, not just for fairness, but to simulate a resource-constrained environment where computational resources are limited.\n",
    "- **Why it matters**: Faster pipelines are more practical and scalable, especially in environments with strict performance or resource limitations."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "939cc7b1-754d-4802-bb1f-e5bf49629971",
   "metadata": {},
   "source": [
    "#### **3. Memory Usage:**"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "588c6701-d795-49a5-a9a8-cf9ee65f5778",
   "metadata": {},
   "source": [
    "- **Peak Memory Usage:** Maximum memory used during pipeline execution.\n",
    "- **Average CPU Usage:** Percentage of a single core used throughout the process.\n",
    "- **Why it matters:** CubeSat will use a processor with limited memory, and the memory utilization must fit within available memory."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3930ec42-0d83-4234-aef8-2963a8095363",
   "metadata": {},
   "source": [
    "#### **4. Algorithm Code Size:**"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ddd8bd34-0222-4875-9ce0-d72fd4462e49",
   "metadata": {},
   "source": [
    "- The total size of your pipeline, including the serialized model and preprocessing function, measured in megabytes (MB).\n",
    "- Algorithm code must fit within a limited available memory"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ce14090f-da90-47ba-8991-ecc7b5420051",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "77219ac5-e71c-4fb1-a1c1-6e0f73d6e4cf",
   "metadata": {},
   "source": [
    "#### Import test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "a44621c4-1dde-4eee-b979-1e20e011bbdd",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "# first let us load the testing data\n",
    "test_images = np.load('data/test_images.npy')      # Load image test data\n",
    "test_labels = np.load('data/test_labels.npy')      # Load label test data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f0a97f81-69b5-4edc-8cb7-8217696f3a73",
   "metadata": {},
   "source": [
    "**Note:** You can also use this notebook to evaluate trained models on the validation set."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "558ff016-bb2c-423e-8fac-8f8b27828219",
   "metadata": {},
   "source": [
    "#### Import Evaluation function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "ee6cb903-25be-4b1b-bd07-fbf505be9f16",
   "metadata": {},
   "outputs": [],
   "source": [
    "from source.evaluate import evaluate_pipeline # A built-in function to evaluate a given ML pipeline by preprocessing, predicting, and calculating performance metrics."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5edf9933-3d8d-4bec-a631-96f94c1ce305",
   "metadata": {},
   "source": [
    "**Inputs:**\n",
    "  \n",
    "- **model**: The trained machine learning model to evaluate.\n",
    "- **X_test_raw**: Raw test data that needs to be preprocessed before evaluation.\n",
    "- **y_test**: True labels corresponding to the test data for performance comparison.\n",
    "- **preprocessing_fn**: A function used to preprocess the raw test data.\n",
    "    \n",
    "**Outputs:**\n",
    "\n",
    "- **metrics**: A dictionary containing various evaluation metrics like accuracy, F1 score, evaluation time, memory usage, CPU usage, and algorithm code size.\n",
    "- **confusion matrix**"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2b2f8ba0-5278-4454-906c-254c9a1ffa28",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f922159c-a415-4678-96d5-fbbf0372a0ad",
   "metadata": {},
   "source": [
    "### 2. ML: Evaluation (3rd notebook)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "034b7805-3e93-46ea-9c77-e691d5d66d87",
   "metadata": {},
   "source": [
    "#### Preprocessing (Testing data)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "54ac8086-8d43-498d-a9b1-c2960c6f28f2",
   "metadata": {},
   "source": [
    "Applying consistent preprocessing to both training and testing data ensures accurate predictions, prevents mismatches, and improves model evaluation. Combining all steps into one function enhances consistency, tracks execution time, and optimizes resource usage."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "9926fa8c-e654-4556-8e0a-c07d4ab8bd30",
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocessing_fn_ML(X): # We apply the same pre-processing steps implemented in Notebook 3.\n",
    "    from skimage.color import rgb2gray\n",
    "    from skimage.transform import resize\n",
    "    \n",
    "    # Normalize the data to [0, 1]\n",
    "    X_pre = X.astype('float32') / 255.0\n",
    "    \n",
    "    # Convert to grayscale\n",
    "    X_pre = np.array([rgb2gray(image) for image in X_pre])\n",
    "    \n",
    "    # Resize images to 64x64 pixels\n",
    "    X_pre = np.array([resize(image, (64, 64), anti_aliasing=True) for image in X_pre])\n",
    "    \n",
    "    # Flatten the images\n",
    "    num_samples = X_pre.shape[0]\n",
    "    X_pre = X_pre.reshape(num_samples, -1)\n",
    "    \n",
    "    return X_pre"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9456d318-101f-47e5-8a33-d27d3ed5a793",
   "metadata": {},
   "source": [
    "#### Import ML model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "42463e21-4df8-4620-861b-7b3cda95ff11",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "# Load the ml model from the 3rd notebook\n",
    "with open('models/sgd_model.pkl', 'rb') as file:\n",
    "    sgd_model = pickle.load(file)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "14307ec6-396a-4ec6-ad20-6ce395f4e73f",
   "metadata": {},
   "source": [
    "#### ML Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "ac5072ad-0d23-416d-8355-f584057d48a9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "### Evaluation Metrics ###\n",
      "\n",
      "Evaluation Time:       35.55 seconds (The time it took for the pipeline to preprocess data and make predictions.)\n",
      "Peak Memory Usage:     18535.05 MB (The maximum memory used during evaluation.)\n",
      "Average CPU Usage:     94.19 % (The % shows how much of one CPU core was used during the evaluation.)\n",
      "Algorithm code size:         0.08 MB (The size of the trained model and preprocessing function.)\n",
      "Accuracy:              0.452 (The percentage of correctly classified samples.)\n",
      "F1 Score:              0.387 (A balance of precision and recall, useful for imbalanced datasets.)\n",
      "\n",
      "### Confusion Matrix ###\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x700 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Assuming you have:\n",
    "# - A trained model named like 'lr_model'\n",
    "# - Raw test data 'X_test_raw'\n",
    "# - True labels 'y_test'\n",
    "# - All pre-processing methods gathered in one function\n",
    "\n",
    "\n",
    "# Evaluate the pipeline\n",
    "metrics = evaluate_pipeline(sgd_model, test_images, test_labels, preprocessing_fn_ML)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8b2ba278-d0c4-4f1f-80a9-b7d02ff31f82",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3512bccf-17e5-4399-8409-edc055906191",
   "metadata": {},
   "source": [
    "### 3. CubeSatNet_CNN Evaluation"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2149f017-0a59-468e-bdcd-188611639bc7",
   "metadata": {},
   "source": [
    "#### preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "c70cc940-e497-4fb6-b86e-801d552a084c",
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocessing_fn_CNN(X):  # we did not use any preprocessing in notebook 4\n",
    "\n",
    "    return X"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c13bf5a4-2e2d-49da-8369-42d86f654bea",
   "metadata": {},
   "source": [
    "#### Import CNN model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "75f6a7ac-3fc2-4623-bb3a-809c183c0100",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2025-01-08 16:25:19.067049: I external/local_xla/xla/tsl/cuda/cudart_stub.cc:32] Could not find cuda drivers on your machine, GPU will not be used.\n",
      "2025-01-08 16:25:20.298813: I external/local_xla/xla/tsl/cuda/cudart_stub.cc:32] Could not find cuda drivers on your machine, GPU will not be used.\n",
      "2025-01-08 16:25:20.643078: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:485] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered\n",
      "2025-01-08 16:25:21.398004: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:8454] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered\n",
      "2025-01-08 16:25:21.598407: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1452] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered\n",
      "2025-01-08 16:25:22.310910: I tensorflow/core/platform/cpu_feature_guard.cc:210] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
      "To enable the following instructions: AVX2 AVX512F FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2025-01-08 16:26:22.046561: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n"
     ]
    }
   ],
   "source": [
    "from keras.utils import to_categorical\n",
    "test_labels = to_categorical(test_labels, num_classes=5)\n",
    "\n",
    "# Load the CNN model from the 4th notebook\n",
    "with open('models/cnn_model.pkl', 'rb') as file:\n",
    "    cnn_model = pickle.load(file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7b7d2d83-debb-4c32-a3ca-59237563b89a",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "6537c79e-14cb-495c-adbf-3d6e6d86ca53",
   "metadata": {},
   "source": [
    "#### Evaluate the CNN pipeline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "32dc05ff-a71b-4cb4-a991-481e5083cdd1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[1m102/102\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m234s\u001b[0m 2s/step\n",
      "\n",
      "### Evaluation Metrics ###\n",
      "\n",
      "Evaluation Time:       265.27 seconds (The time it took for the pipeline to preprocess data and make predictions.)\n",
      "Peak Memory Usage:     8455.79 MB (The maximum memory used during evaluation.)\n",
      "Average CPU Usage:     89.29 % (The % shows how much of one CPU core was used during the evaluation.)\n",
      "Algorithm code size:         1.16 MB (The size of the trained model and preprocessing function.)\n",
      "Accuracy:              0.998 (The percentage of correctly classified samples.)\n",
      "F1 Score:              0.998 (A balance of precision and recall, useful for imbalanced datasets.)\n",
      "\n",
      "### Confusion Matrix ###\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x700 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Evaluate the pipeline\n",
    "metrics = evaluate_pipeline(cnn_model, test_images, test_labels, preprocessing_fn_CNN)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2eaa2a30-1929-42cc-b456-d2ccbacc21a1",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e310d186-6a57-4595-8d88-65f8fd1c95a3",
   "metadata": {},
   "source": [
    "##### **⚠️ Freeing up Space**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "6fa09b17-a600-4ebf-9aec-66920457cbfb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Data and models removed from memory.\n"
     ]
    }
   ],
   "source": [
    "import gc\n",
    "\n",
    "# Remove the data from memory\n",
    "del sgd_model, cnn_model, test_images, test_labels\n",
    "\n",
    "# Force garbage collection to free up memory\n",
    "gc.collect()\n",
    "\n",
    "# Clear the input/output cache\n",
    "print(\"Data and models removed from memory.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5852146c-2579-4eff-a795-7002b09cec3c",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a4d151f8-8749-49ab-bdc6-0da5617f3b97",
   "metadata": {},
   "source": [
    "**Note:**\n",
    "\n",
    "1.\tYou can also use the `evaluate_pipeline` function to evaluate your validation set data.\n",
    "2.\tPlease do not modify the `evaluate_pipeline` function.\n",
    "\n",
    "Best of luck with the hackathon challenge! 🎉 We look forward to seeing your final results and pipelines! 🚀"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "CubeHackKer",
   "language": "python",
   "name": "cubehackker"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}