Preprocessing.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt"
   ],
   "metadata": {
    "id": "bwXmp_jDmZpr",
    "executionInfo": {
     "status": "ok",
     "timestamp": 1702764882922,
     "user_tz": -60,
     "elapsed": 1751,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "bwXmp_jDmZpr",
   "execution_count": 1,
   "outputs": []
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "outputs": [],
   "source": [
    "# connect, give access, and \"download\" your drive on the google VM\n",
    "#from google.colab import drive\n",
    "#drive.mount('/content/drive')"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "source": [
    "# change to your path to the dataset in drive (create a shortcut of shared ML_project in \"MyDrive\")\n",
    "#your_drive_path = '/content/drive/MyDrive/ML_project/dataset.parquet'\n",
    "local_path = 'dataset.parquet'"
   ],
   "metadata": {
    "id": "f6F2feg8mH_e",
    "executionInfo": {
     "status": "ok",
     "timestamp": 1702764884690,
     "user_tz": -60,
     "elapsed": 7,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "f6F2feg8mH_e",
   "execution_count": 2,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "df = pd.read_parquet(local_path, engine='pyarrow')"
   ],
   "metadata": {
    "id": "Idjil-Yqq4pI",
    "executionInfo": {
     "status": "ok",
     "timestamp": 1702764890090,
     "user_tz": -60,
     "elapsed": 5406,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "Idjil-Yqq4pI",
   "execution_count": 3,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": [
    "# Initial preprocessing"
   ],
   "metadata": {
    "id": "rAci_wZ0xsHd"
   },
   "id": "rAci_wZ0xsHd"
  },
  {
   "cell_type": "markdown",
   "source": [
    "We create a video_id which identifies each video.\n"
   ],
   "metadata": {
    "id": "9QDm2Egq2FmJ"
   },
   "id": "9QDm2Egq2FmJ"
  },
  {
   "cell_type": "code",
   "source": [
    "# Find the rows where time is 0\n",
    "start_of_video = df['time(s)'] == 0\n",
    "\n",
    "# Use cumsum to create a unique video_id for each video\n",
    "df['video_id'] = start_of_video.cumsum() - 1\n",
    "\n",
    "# Forward-fill the video_id to all rows within the same video series\n",
    "df['video_id'] = df['video_id'].ffill(downcast='int')\n",
    "\n",
    "# Reorder columns, moving 'video_id' to the first position\n",
    "df = df[['video_id'] + [col for col in df.columns if col != 'video_id']]"
   ],
   "metadata": {
    "id": "jYn-Sl9Vxl8o",
    "executionInfo": {
     "status": "ok",
     "timestamp": 1702764891308,
     "user_tz": -60,
     "elapsed": 1257,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "jYn-Sl9Vxl8o",
   "execution_count": 4,
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\41794\\AppData\\Local\\Temp\\ipykernel_18768\\2335139889.py:8: FutureWarning: The 'downcast' keyword in ffill is deprecated and will be removed in a future version. Use res.infer_objects(copy=False) to infer non-object dtype, or pd.to_numeric with the 'downcast' keyword to downcast numeric results.\n",
      "  df['video_id'] = df['video_id'].ffill(downcast='int')\n"
     ]
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "We then onehot encode exercises."
   ],
   "metadata": {
    "id": "EA4K6sBjowBg"
   },
   "id": "EA4K6sBjowBg"
  },
  {
   "cell_type": "code",
   "source": [
    "one_hot_encoded_exercises = pd.get_dummies(df['Exercise'], prefix='encoded_exo')\n",
    "\n",
    "# Concatenate the new one-hot encoded columns to the original DataFrame\n",
    "df = pd.concat([df, one_hot_encoded_exercises], axis=1)"
   ],
   "metadata": {
    "id": "jgaPtHqPEOiJ",
    "executionInfo": {
     "status": "ok",
     "timestamp": 1702764892819,
     "user_tz": -60,
     "elapsed": 1515,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "jgaPtHqPEOiJ",
   "execution_count": 5,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": [
    "We first drop the first 2 seconds and the last second of each video, because they approximately correspond to a reference pose and clapping."
   ],
   "metadata": {
    "id": "1fsLaMoF0iRa"
   },
   "id": "1fsLaMoF0iRa"
  },
  {
   "cell_type": "code",
   "source": [
    "# Drop the first 60 frames and the last 30 frames for each video\n",
    "df = df.groupby('video_id').apply(lambda group: group.iloc[60:-30]).reset_index(drop=True)"
   ],
   "metadata": {
    "id": "OxkkKcP5yFDw",
    "executionInfo": {
     "status": "ok",
     "timestamp": 1702764895948,
     "user_tz": -60,
     "elapsed": 3135,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "OxkkKcP5yFDw",
   "execution_count": 6,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "source": [
    "We will now handle missing values. We will crop parts of videos that have more than 20 (obtained by cross validation) missing continuous frames. Then we will perform linear interpolation to estimate missing joint positions that consist of less than 20 continuous frames."
   ],
   "metadata": {
    "id": "UYsnyWK807MU"
   },
   "id": "UYsnyWK807MU"
  },
  {
   "cell_type": "code",
   "source": [
    "nan_mask = df.isna().any(axis = 1)\n",
    "# Get the indices of the rows with NaN values\n",
    "nan_indices = [i for i, has_nan in enumerate(nan_mask) if has_nan]\n",
    "\n",
    "th = 20\n",
    "current_sequence_length = 1\n",
    "ids = []\n",
    "tmp = []\n",
    "\n",
    "# Get indices that have more than 20 continuous missing frames\n",
    "for i in range(1, len(nan_indices)-1):\n",
    "    tmp.append(nan_indices[i])\n",
    "    if nan_indices[i] == nan_indices[i + 1] - 1:\n",
    "        current_sequence_length += 1\n",
    "    else:\n",
    "        if current_sequence_length > th:\n",
    "            ids.append(tmp)\n",
    "        tmp = []\n",
    "        current_sequence_length = 1\n",
    "ids = np.concatenate(ids)\n",
    "\n",
    "# Drop rows based on iloc indices\n",
    "df = df.drop(ids)\n",
    "\n",
    "# Reset index if needed\n",
    "df = df.reset_index(drop=True)"
   ],
   "metadata": {
    "id": "qZoe-CFO8nyT",
    "executionInfo": {
     "status": "ok",
     "timestamp": 1702764898768,
     "user_tz": -60,
     "elapsed": 2874,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "qZoe-CFO8nyT",
   "execution_count": 7,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "#Interpolating linearly for the missing frames\n",
    "df = df.interpolate(method='linear', axis=0, limit_direction='both')"
   ],
   "metadata": {
    "id": "gBZCEs3_xRfe",
    "executionInfo": {
     "status": "ok",
     "timestamp": 1702764907207,
     "user_tz": -60,
     "elapsed": 8449,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "gBZCEs3_xRfe",
   "execution_count": 8,
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\41794\\AppData\\Local\\Temp\\ipykernel_18768\\3724438224.py:2: FutureWarning: DataFrame.interpolate with object dtype is deprecated and will raise in a future version. Call obj.infer_objects(copy=False) before interpolating instead.\n",
      "  df = df.interpolate(method='linear', axis=0, limit_direction='both')\n"
     ]
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "# Train / test split"
   ],
   "metadata": {
    "id": "pNaGmIbZRJYF"
   },
   "id": "pNaGmIbZRJYF"
  },
  {
   "cell_type": "markdown",
   "source": [
    "We now split the dataset into two dataframes, containing the train and test sets. We select 5 participants that will be in the test set, while others will be in the train set."
   ],
   "metadata": {
    "id": "6sHwfP6zpSc7"
   },
   "id": "6sHwfP6zpSc7"
  },
  {
   "cell_type": "code",
   "source": [
    "# Combine 'Set' and 'Exercise' columns to create a new column for grouping\n",
    "df['Combined_Labels'] = df['Set'].astype(str) + '_' + df['Exercise'].astype(str)\n",
    "\n",
    "# Select participants that will be part of the test set\n",
    "test_participants = ['P04','P05','P09','P11','P12']\n",
    "\n",
    "# Split into train and test sets\n",
    "train_set = df[~df['Participant'].isin(test_participants)]\n",
    "test_set = df[df['Participant'].isin(test_participants)]\n",
    "\n",
    "# Check that both sets contain a maximum amount of labels\n",
    "unique_labels_train = train_set['Combined_Labels'].nunique()\n",
    "unique_labels_test = test_set['Combined_Labels'].nunique()\n",
    "\n",
    "print(\"Train set contains \",unique_labels_train,\"/38 labels.\")\n",
    "print(\"Test set contains \",unique_labels_test,\"/38 labels.\")\n",
    "\n",
    "# Drop the temporary 'Combined_Labels' column\n",
    "train_set = train_set.drop('Combined_Labels', axis=1)\n",
    "test_set = test_set.drop('Combined_Labels', axis=1)"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "zn9QqLY_WjpA",
    "executionInfo": {
     "status": "ok",
     "timestamp": 1702764910122,
     "user_tz": -60,
     "elapsed": 2938,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    },
    "outputId": "2f75d6a6-f625-46b8-fda6-41722ccf0582"
   },
   "id": "zn9QqLY_WjpA",
   "execution_count": 9,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train set contains  38 /38 labels.\n",
      "Test set contains  37 /38 labels.\n"
     ]
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "# Data Augmentation"
   ],
   "metadata": {
    "id": "2tkNYGEqRCH_"
   },
   "id": "2tkNYGEqRCH_"
  },
  {
   "cell_type": "markdown",
   "source": [
    "Let's look at the distribution of our train set"
   ],
   "metadata": {
    "id": "pkQ325e4Uh5R"
   },
   "id": "pkQ325e4Uh5R"
  },
  {
   "cell_type": "code",
   "source": [
    "# Aggregate the data at the video level\n",
    "video_aggregated = train_set.groupby(['video_id', 'Exercise', 'Set']).size().reset_index(name='Count')\n",
    "\n",
    "# Plot the distribution of Sets within each Exercise\n",
    "plt.figure(figsize=(8, 4))\n",
    "sns.countplot(x='Exercise', hue='Set', data=video_aggregated)\n",
    "\n",
    "# Reduce the size of the legend\n",
    "plt.legend(prop={'size': 6})\n",
    "\n",
    "plt.title('Distribution of Sets within each Exercise in our train set')\n",
    "plt.show()"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 410
    },
    "id": "z0elWWtyWOgL",
    "executionInfo": {
     "status": "ok",
     "timestamp": 1702764911532,
     "user_tz": -60,
     "elapsed": 1424,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    },
    "outputId": "fe783781-cd15-4962-ad3e-f8139d528e58"
   },
   "id": "z0elWWtyWOgL",
   "execution_count": 10,
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\41794\\anaconda3\\envs\\ada\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "C:\\Users\\41794\\anaconda3\\envs\\ada\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "C:\\Users\\41794\\anaconda3\\envs\\ada\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "C:\\Users\\41794\\anaconda3\\envs\\ada\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n"
     ]
    },
    {
     "data": {
      "text/plain": "<Figure size 800x400 with 1 Axes>",
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "We can see that our train set is heavily unbalanced. We will thus, balance our dataset by randomly sampling a video of a certain type (Exercise,Set), duplicate it, add gaussian noise to it, and append it to our train set. This will also make the model more robust to noise."
   ],
   "metadata": {
    "id": "pbbwdLQkTiRa"
   },
   "id": "pbbwdLQkTiRa"
  },
  {
   "cell_type": "code",
   "source": [
    "def add_gaussian_noise_to_series(series_df, video_id, noise_std):\n",
    "    # Extract the video series based on the provided video_id\n",
    "    video_series = series_df[series_df['video_id'] == video_id].copy()\n",
    "\n",
    "    # Add Gaussian noise to the joint positions\n",
    "    for col in video_series.columns[6:105]:\n",
    "        noise = np.random.normal(0, noise_std, len(video_series))\n",
    "        video_series[col] += noise\n",
    "\n",
    "    # Clip the joint positions to be within the -1 to 1 range\n",
    "    video_series.iloc[:, 6:105] = np.clip(video_series.iloc[:, 6:105], -1, 1)\n",
    "\n",
    "    # Increment the series_id for the new video\n",
    "    video_series['video_id'] = series_df['video_id'].max() + 1\n",
    "\n",
    "    # Append the new video to the DataFrame\n",
    "    return pd.concat([series_df, video_series], ignore_index=True)"
   ],
   "metadata": {
    "id": "HdpLpCmTVMEm",
    "executionInfo": {
     "status": "ok",
     "timestamp": 1702764911532,
     "user_tz": -60,
     "elapsed": 10,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "HdpLpCmTVMEm",
   "execution_count": 12,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "def augment_videos(combination, df, desired_count):\n",
    "    # Extract exercise and set values from the combination\n",
    "    exercise, set_value = combination\n",
    "\n",
    "    # Count the current number of unique videos for the given combination of exercise and set\n",
    "    current_count = len(df[(df['Exercise'] == exercise) & (df['Set'] == set_value)]['video_id'].unique())\n",
    "\n",
    "    # Calculate the number of videos to add to reach the desired count\n",
    "    videos_to_add = desired_count - current_count\n",
    "\n",
    "    # Loop to add videos with Gaussian noise\n",
    "    for _ in range(videos_to_add):\n",
    "        # Set the standard deviation for Gaussian noise\n",
    "        noise_std = 0.1\n",
    "\n",
    "        # Sample a random video_id from the given combination of exercise and set\n",
    "        sampled_video_id = df[(df['Exercise'] == exercise) & (df['Set'] == set_value)]['video_id'].sample().values[0]\n",
    "\n",
    "        # Apply the function to add Gaussian noise to the sampled video\n",
    "        df = add_gaussian_noise_to_series(df, sampled_video_id, noise_std)\n",
    "\n",
    "    return df"
   ],
   "metadata": {
    "id": "7L3-RGPX_Hj4",
    "executionInfo": {
     "status": "ok",
     "timestamp": 1702764911533,
     "user_tz": -60,
     "elapsed": 9,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "7L3-RGPX_Hj4",
   "execution_count": 13,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "# Define the desired number of videos per combination\n",
    "desired_videos_per_combination = 85\n",
    "\n",
    "# Group by \"Exercise\" and \"Set\" and count the number of unique videos in each group\n",
    "video_counts = train_set.groupby(['Exercise', 'Set'])['video_id'].nunique()\n",
    "\n",
    "# Identify combinations that need augmentation\n",
    "combinations_to_augment = video_counts[video_counts < desired_videos_per_combination].index\n",
    "\n",
    "\n",
    "# Augment videos for combinations that need it\n",
    "for combination in combinations_to_augment:\n",
    "    train_set = augment_videos(combination, train_set, desired_videos_per_combination)"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 356
    },
    "id": "bi4rPHoR9Zoy",
    "executionInfo": {
     "status": "error",
     "timestamp": 1702765040649,
     "user_tz": -60,
     "elapsed": 129125,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    },
    "outputId": "06aa4511-13d1-4dd6-ef91-99f702b1b324"
   },
   "id": "bi4rPHoR9Zoy",
   "execution_count": 14,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "# Verify the result\n",
    "\n",
    "# Aggregate the data at the video level\n",
    "video_aggregated = train_set.groupby(['video_id', 'Exercise', 'Set']).size().reset_index(name='Count')\n",
    "\n",
    "# Plot the distribution of Sets within each Exercise\n",
    "plt.figure(figsize=(8, 4))\n",
    "sns.countplot(x='Exercise', hue='Set', data=video_aggregated)\n",
    "\n",
    "# Reduce the size of the legend\n",
    "plt.legend(prop={'size': 6})\n",
    "\n",
    "plt.title('Distribution of Sets within Each Exercise in our train set')\n",
    "plt.show()"
   ],
   "metadata": {
    "id": "wZ1oCYhkPCyI",
    "executionInfo": {
     "status": "aborted",
     "timestamp": 1702758817438,
     "user_tz": -60,
     "elapsed": 31,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "wZ1oCYhkPCyI",
   "execution_count": 15,
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\41794\\anaconda3\\envs\\ada\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "C:\\Users\\41794\\anaconda3\\envs\\ada\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "C:\\Users\\41794\\anaconda3\\envs\\ada\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "C:\\Users\\41794\\anaconda3\\envs\\ada\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n"
     ]
    },
    {
     "data": {
      "text/plain": "<Figure size 800x400 with 1 Axes>",
      "image/png": "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"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "The train set it now balanced."
   ],
   "metadata": {
    "id": "yP8uVxeCcCb9"
   },
   "id": "yP8uVxeCcCb9"
  },
  {
   "cell_type": "code",
   "source": [
    "test_path = 'test_set.parquet'\n",
    "test_set.to_parquet(test_path, index=True)"
   ],
   "metadata": {
    "id": "GVQbgQmQNt83",
    "executionInfo": {
     "status": "aborted",
     "timestamp": 1702758817440,
     "user_tz": -60,
     "elapsed": 32,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "GVQbgQmQNt83",
   "execution_count": 16,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "train_path = 'train_set_augmented.parquet'\n",
    "train_set.to_parquet(train_path, index=True)"
   ],
   "metadata": {
    "id": "fBwnrDSIOtmT",
    "executionInfo": {
     "status": "aborted",
     "timestamp": 1702758817441,
     "user_tz": -60,
     "elapsed": 33,
     "user": {
      "displayName": "Garik AMV",
      "userId": "17416187305741627565"
     }
    }
   },
   "id": "fBwnrDSIOtmT",
   "execution_count": 11,
   "outputs": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "name": "conda-env-ada-py",
   "language": "python",
   "display_name": "Python [conda env:ada] *"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  },
  "colab": {
   "provenance": [
    {
     "file_id": "https://github.com/RamiATASSI/ML_project2/blob/main/Tests.ipynb",
     "timestamp": 1701879533842
    }
   ]
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}