Added time in seconds and minutes (not just quarter seconds)

birth-outcomes · Nov 27, 2023 · cec746f · cec746f
1 parent aba497b
commit cec746f
Show file tree

Hide file tree

Showing 553 changed files with 9,825,764 additions and 9,825,734 deletions.
diff --git a/01_create_csv_database.ipynb b/01_create_csv_database.ipynb
@@ -8,7 +8,7 @@
     "\n",
     "This converts the `.dat` and `.hea` files in the `data/` folder into `.csv` files stored in the `data_csv/` folder.\n",
     "\n",
-    "Source: https://github.com/fabiom91/CTU-CHB_Physionet.org/blob/master/create_csv_database.ipynb"
+    "Code adapted from: https://github.com/fabiom91/CTU-CHB_Physionet.org/blob/master/create_csv_database.ipynb"
    ]
   },
   {
@@ -24,10 +24,11 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "import pandas as pd\n",
-    "import wfdb\n",
+    "import math\n",
     "from os import listdir\n",
+    "import pandas as pd\n",
     "from tqdm import tqdm\n",
+    "import wfdb\n",
     "\n",
     "import warnings\n",
     "warnings.filterwarnings(\"ignore\")"
@@ -62,15 +63,44 @@
     "    return rec_list\n",
     "\n",
     "\n",
+    "def consec_repeat(repeats, max_value, len):\n",
+    "    '''\n",
+    "    Creates list with consecutive numbers, where each number repeats a specified\n",
+    "    amount of times\n",
+    "    Inputs:\n",
+    "    - repeats - int, number of times to repeat each number\n",
+    "    - max_value - int, maximum number to reach\n",
+    "    - len - int, length of list (as may be to a quarter second so not 4 end rep)\n",
+    "    '''\n",
+    "    # Round up max_value to nearest integer if required\n",
+    "    max_value = math.ceil(max_value)\n",
+    "    # Create list of repeated consecutive numbers\n",
+    "    list = [x//repeats for x in range((max_value+1)*repeats)]\n",
+    "    # Trim list to desired length (to deal with odd quarter seconds and with\n",
+    "    # the zero-based indexing (whilst max_value will be one higher))\n",
+    "    list = list[:len]\n",
+    "    return(list)\n",
+    "\n",
+    "\n",
     "def create_signals_database(rec):\n",
     "    '''\n",
-    "    Read the signal files (with FHR and UC) and save to csv\n",
+    "    Read the signal files (with FHR and UC), add tme in seconds and minutes,\n",
+    "    and save to csv file.\n",
     "    Inputs:\n",
     "    - rec - name of record (e.g. '1347')\n",
     "    '''\n",
+    "    # Read signal file and save to dataframe\n",
     "    sample = wfdb.rdsamp(\"data/%s\" % rec)\n",
     "    df = pd.DataFrame(sample[0], columns=['FHR','UC'])\n",
     "    df.index.name = 'quarter_second'\n",
+    "    # Find the length of the record in quarter seconds, seconds and minutes\n",
+    "    q_sec = len(df.index)\n",
+    "    sec = q_sec/4\n",
+    "    min = sec/60\n",
+    "    # Add column with time in seconds and minutes\n",
+    "    df['second'] = consec_repeat(repeats=4, max_value=int(sec), len=q_sec)\n",
+    "    df['minute'] = consec_repeat(repeats=4*60, max_value=int(min), len=q_sec)\n",
+    "    # Save to csv file\n",
     "    df.to_csv('data_csv/%s.csv' % rec)\n",
     "\n",
     "\n",
@@ -130,7 +160,7 @@
      "name": "stderr",
      "output_type": "stream",
      "text": [
-      "100%|██████████| 1104/1104 [00:50<00:00, 21.65it/s]"
+      "100%|██████████| 1104/1104 [01:12<00:00, 15.27it/s]"
      ]
     },
     {