Skip to content

Instantly share code, notes, and snippets.

@AndreFCruz

AndreFCruz/testing-nan-bug.ipynb

Created May 5, 2024 12:02
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save AndreFCruz/97fd5aa5aff4ddd610572eeadc548fa2 to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save AndreFCruz/97fd5aa5aff4ddd610572eeadc548fa2 to your computer and use it in GitHub Desktop.
Download ZIP
Investigating a NaN bug on folktables
Display the source blob
Display the rendered blob
Raw
testing-nan-bug.ipynb
{
"cells": [
{
"cell_type": "markdown",
"id": "aa986947-ff09-4041-b727-6abb9fc29c99",
"metadata": {},
"source": [
"## Preprocessing"
]
},
{
"cell_type": "code",
"execution_count": 1,
"id": "18f8a0c4-0187-4947-bbb6-8f9b3a759cc7",
"metadata": {},
"outputs": [],
"source": [
"import sys\n",
"import logging\n",
"from pathlib import Path\n",
"\n",
"import pandas as pd\n",
"import numpy as np\n",
"from folktables import ACSDataSource"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "9867c806-3527-42e8-a8a1-42148dfa00ea",
"metadata": {},
"outputs": [],
"source": [
"root_dir = Path(\"~\").expanduser()\n",
"data_dir = root_dir / \"data\" / \"folktables\"\n",
"data_dir.mkdir(parents=True, exist_ok=True)"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "49cc60f1-eb09-4524-9b9e-6faf62b5560a",
"metadata": {},
"outputs": [],
"source": [
"# download 2018 ACS data\n",
"from folktables.load_acs import state_list\n",
"\n",
"data_source = ACSDataSource(\n",
" survey_year='2018', horizon='1-Year', survey='person',\n",
" root_dir=str(data_dir),\n",
")"
]
},
{
"cell_type": "code",
"execution_count": 4,
"id": "a2095dc3-47b0-481e-990b-ea0303ce877c",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"(3236107, 286)"
]
},
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# data is 3236107 rows x 286 columns\n",
"acs_data = data_source.get_data(states=state_list, download=True) # use download=True if not yet downloaded\n",
"acs_data.shape"
]
},
{
"cell_type": "markdown",
"id": "facbd077-89f6-46d5-b8ac-5404da0e7571",
"metadata": {},
"source": [
"---\n",
"## Search for NaNs"
]
},
{
"cell_type": "code",
"execution_count": 5,
"id": "ec607408-29e3-4696-bf13-48fc1a017acf",
"metadata": {},
"outputs": [
{
"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>RT</th>\n",
" <th>SERIALNO</th>\n",
" <th>DIVISION</th>\n",
" <th>SPORDER</th>\n",
" <th>PUMA</th>\n",
" <th>REGION</th>\n",
" <th>ST</th>\n",
" <th>ADJINC</th>\n",
" <th>PWGTP</th>\n",
" <th>AGEP</th>\n",
" <th>...</th>\n",
" <th>PWGTP71</th>\n",
" <th>PWGTP72</th>\n",
" <th>PWGTP73</th>\n",
" <th>PWGTP74</th>\n",
" <th>PWGTP75</th>\n",
" <th>PWGTP76</th>\n",
" <th>PWGTP77</th>\n",
" <th>PWGTP78</th>\n",
" <th>PWGTP79</th>\n",
" <th>PWGTP80</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
" <td>P</td>\n",
" <td>2018GQ0000049</td>\n",
" <td>6</td>\n",
" <td>1</td>\n",
" <td>1600</td>\n",
" <td>3</td>\n",
" <td>1</td>\n",
" <td>1013097</td>\n",
" <td>75</td>\n",
" <td>19</td>\n",
" <td>...</td>\n",
" <td>140</td>\n",
" <td>74</td>\n",
" <td>73</td>\n",
" <td>7</td>\n",
" <td>76</td>\n",
" <td>75</td>\n",
" <td>80</td>\n",
" <td>74</td>\n",
" <td>7</td>\n",
" <td>72</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>P</td>\n",
" <td>2018GQ0000058</td>\n",
" <td>6</td>\n",
" <td>1</td>\n",
" <td>1900</td>\n",
" <td>3</td>\n",
" <td>1</td>\n",
" <td>1013097</td>\n",
" <td>75</td>\n",
" <td>18</td>\n",
" <td>...</td>\n",
" <td>76</td>\n",
" <td>78</td>\n",
" <td>7</td>\n",
" <td>76</td>\n",
" <td>80</td>\n",
" <td>78</td>\n",
" <td>7</td>\n",
" <td>147</td>\n",
" <td>150</td>\n",
" <td>75</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>P</td>\n",
" <td>2018GQ0000219</td>\n",
" <td>6</td>\n",
" <td>1</td>\n",
" <td>2000</td>\n",
" <td>3</td>\n",
" <td>1</td>\n",
" <td>1013097</td>\n",
" <td>118</td>\n",
" <td>53</td>\n",
" <td>...</td>\n",
" <td>117</td>\n",
" <td>121</td>\n",
" <td>123</td>\n",
" <td>205</td>\n",
" <td>208</td>\n",
" <td>218</td>\n",
" <td>120</td>\n",
" <td>19</td>\n",
" <td>123</td>\n",
" <td>18</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>P</td>\n",
" <td>2018GQ0000246</td>\n",
" <td>6</td>\n",
" <td>1</td>\n",
" <td>2400</td>\n",
" <td>3</td>\n",
" <td>1</td>\n",
" <td>1013097</td>\n",
" <td>43</td>\n",
" <td>28</td>\n",
" <td>...</td>\n",
" <td>43</td>\n",
" <td>76</td>\n",
" <td>79</td>\n",
" <td>77</td>\n",
" <td>80</td>\n",
" <td>44</td>\n",
" <td>46</td>\n",
" <td>82</td>\n",
" <td>81</td>\n",
" <td>8</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>P</td>\n",
" <td>2018GQ0000251</td>\n",
" <td>6</td>\n",
" <td>1</td>\n",
" <td>2701</td>\n",
" <td>3</td>\n",
" <td>1</td>\n",
" <td>1013097</td>\n",
" <td>16</td>\n",
" <td>25</td>\n",
" <td>...</td>\n",
" <td>4</td>\n",
" <td>2</td>\n",
" <td>29</td>\n",
" <td>17</td>\n",
" <td>15</td>\n",
" <td>28</td>\n",
" <td>17</td>\n",
" <td>30</td>\n",
" <td>15</td>\n",
" <td>1</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"<p>5 rows × 286 columns</p>\n",
"</div>"
],
"text/plain": [
" RT SERIALNO DIVISION SPORDER PUMA REGION ST ADJINC PWGTP \\\n",
"0 P 2018GQ0000049 6 1 1600 3 1 1013097 75 \n",
"1 P 2018GQ0000058 6 1 1900 3 1 1013097 75 \n",
"2 P 2018GQ0000219 6 1 2000 3 1 1013097 118 \n",
"3 P 2018GQ0000246 6 1 2400 3 1 1013097 43 \n",
"4 P 2018GQ0000251 6 1 2701 3 1 1013097 16 \n",
"\n",
" AGEP ... PWGTP71 PWGTP72 PWGTP73 PWGTP74 PWGTP75 PWGTP76 PWGTP77 \\\n",
"0 19 ... 140 74 73 7 76 75 80 \n",
"1 18 ... 76 78 7 76 80 78 7 \n",
"2 53 ... 117 121 123 205 208 218 120 \n",
"3 28 ... 43 76 79 77 80 44 46 \n",
"4 25 ... 4 2 29 17 15 28 17 \n",
"\n",
" PWGTP78 PWGTP79 PWGTP80 \n",
"0 74 7 72 \n",
"1 147 150 75 \n",
"2 19 123 18 \n",
"3 82 81 8 \n",
"4 30 15 1 \n",
"\n",
"[5 rows x 286 columns]"
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"acs_data.head()"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "8c7b28fb-3aba-4979-b4b7-a8dafc11f626",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"['RT', 'SERIALNO', 'NAICSP', 'SOCP']"
]
},
"execution_count": 6,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"numeric_cols = [col for col in acs_data.columns if np.issubdtype(acs_data[col].dtype, np.number)]\n",
"acs_data_numeric = acs_data[numeric_cols]\n",
"\n",
"non_numeric_cols = [col for col in acs_data.columns if col not in numeric_cols]\n",
"non_numeric_cols"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "001dcf89-c4a4-4587-bf2b-327c33f5788f",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Column CITWP has 3027081 NaNs (93.5%).\n",
"Column COW has 1322223 NaNs (40.9%).\n",
"Column DDRS has 162925 NaNs (5.0%).\n",
"Column DOUT has 535206 NaNs (16.5%).\n",
"Column DPHY has 162925 NaNs (5.0%).\n",
"Column DRAT has 3185384 NaNs (98.4%).\n",
"Column DRATX has 2986009 NaNs (92.3%).\n",
"Column DREM has 162925 NaNs (5.0%).\n",
"Column ENG has 2649542 NaNs (81.9%).\n",
"Column FER has 2536823 NaNs (78.4%).\n",
"Column GCL has 1126766 NaNs (34.8%).\n",
"Column GCM has 3211594 NaNs (99.2%).\n",
"Column GCR has 3167599 NaNs (97.9%).\n",
"Column INTP has 535206 NaNs (16.5%).\n",
"\t\t\t 2330183 zeros (72.0%).\n",
"Column JWMNP has 1836484 NaNs (56.7%).\n",
"Column JWRIP has 1967271 NaNs (60.8%).\n",
"Column JWTR has 1752768 NaNs (54.2%).\n",
"Column LANX has 162925 NaNs (5.0%).\n",
"Column MARHD has 1341266 NaNs (41.4%).\n",
"Column MARHM has 1341266 NaNs (41.4%).\n",
"Column MARHT has 1341266 NaNs (41.4%).\n",
"Column MARHW has 1341266 NaNs (41.4%).\n",
"Column MARHYP has 1341266 NaNs (41.4%).\n",
"Column MIG has 30520 NaNs (0.9%).\n",
"Column MIL has 614704 NaNs (19.0%).\n",
"Column MLPA has 3021594 NaNs (93.4%).\n",
"\t\t\t 169746 zeros (5.2%).\n",
"Column MLPB has 3021594 NaNs (93.4%).\n",
"\t\t\t 173760 zeros (5.4%).\n",
"Column MLPCD has 3021594 NaNs (93.4%).\n",
"\t\t\t 160608 zeros (5.0%).\n",
"Column MLPE has 3021594 NaNs (93.4%).\n",
"\t\t\t 133691 zeros (4.1%).\n",
"Column MLPFG has 3021594 NaNs (93.4%).\n",
"\t\t\t 186670 zeros (5.8%).\n",
"Column MLPH has 3021594 NaNs (93.4%).\n",
"\t\t\t 197956 zeros (6.1%).\n",
"Column MLPI has 3021594 NaNs (93.4%).\n",
"\t\t\t 212348 zeros (6.6%).\n",
"Column MLPJ has 3021594 NaNs (93.4%).\n",
"\t\t\t 208638 zeros (6.4%).\n",
"Column MLPK has 3021594 NaNs (93.4%).\n",
"\t\t\t 214356 zeros (6.6%).\n",
"Column NWAB has 574865 NaNs (17.8%).\n",
"Column NWAV has 574865 NaNs (17.8%).\n",
"Column NWLA has 574865 NaNs (17.8%).\n",
"Column NWLK has 574865 NaNs (17.8%).\n",
"Column NWRE has 574865 NaNs (17.8%).\n",
"Column OIP has 535206 NaNs (16.5%).\n",
"\t\t\t 2548615 zeros (78.8%).\n",
"Column PAP has 535206 NaNs (16.5%).\n",
"\t\t\t 2666594 zeros (82.4%).\n",
"Column RETP has 535206 NaNs (16.5%).\n",
"\t\t\t 2369699 zeros (73.2%).\n",
"Column SCH has 95465 NaNs (2.9%).\n",
"Column SCHG has 2484712 NaNs (76.8%).\n",
"Column SCHL has 95465 NaNs (2.9%).\n",
"Column SEMP has 535206 NaNs (16.5%).\n",
"\t\t\t 2536197 zeros (78.4%).\n",
"Column SSIP has 535206 NaNs (16.5%).\n",
"\t\t\t 2616173 zeros (80.8%).\n",
"Column SSP has 535206 NaNs (16.5%).\n",
"\t\t\t 2042064 zeros (63.1%).\n",
"Column WAGP has 535206 NaNs (16.5%).\n",
"\t\t\t 1122407 zeros (34.7%).\n",
"Column WKHP has 1557981 NaNs (48.1%).\n",
"Column WKL has 574865 NaNs (17.8%).\n",
"Column WKW has 1557981 NaNs (48.1%).\n",
"Column WRK has 859027 NaNs (26.5%).\n",
"Column YOEP has 2809462 NaNs (86.8%).\n",
"Column DECADE has 2809462 NaNs (86.8%).\n",
"Column DRIVESP has 1967271 NaNs (60.8%).\n",
"Column ESP has 2619012 NaNs (80.9%).\n",
"Column ESR has 574865 NaNs (17.8%).\n",
"Column FOD1P has 2439281 NaNs (75.4%).\n",
"Column FOD2P has 3148321 NaNs (97.3%).\n",
"Column INDP has 1322223 NaNs (40.9%).\n",
"Column JWAP has 1836484 NaNs (56.7%).\n",
"Column JWDP has 1836484 NaNs (56.7%).\n",
"Column LANP has 2649542 NaNs (81.9%).\n",
"Column MIGPUMA has 2809427 NaNs (86.8%).\n",
"Column MIGSP has 2809427 NaNs (86.8%).\n",
"Column MSP has 535206 NaNs (16.5%).\n",
"Column NOP has 2619012 NaNs (80.9%).\n",
"Column OC has 153610 NaNs (4.7%).\n",
"\t\t\t 2513301 zeros (77.7%).\n",
"Column OCCP has 1322223 NaNs (40.9%).\n",
"Column PAOC has 1927168 NaNs (59.6%).\n",
"Column PERNP has 574865 NaNs (17.8%).\n",
"\t\t\t 983842 zeros (30.4%).\n",
"Column PINCP has 535206 NaNs (16.5%).\n",
"\t\t\t 336113 zeros (10.4%).\n",
"Column POVPIP has 135180 NaNs (4.2%).\n",
"\t\t\t 57749 zeros (1.8%).\n",
"Column POWPUMA has 1752768 NaNs (54.2%).\n",
"Column POWSP has 1752768 NaNs (54.2%).\n",
"Column RC has 153610 NaNs (4.7%).\n",
"\t\t\t 2444568 zeros (75.5%).\n",
"Column SCIENGP has 2439281 NaNs (75.4%).\n",
"Column SCIENGRLP has 2439281 NaNs (75.4%).\n",
"Column SFN has 3136282 NaNs (96.9%).\n",
"Column SFR has 3136282 NaNs (96.9%).\n",
"Column VPS has 3021594 NaNs (93.4%).\n",
"Column FHINS3C has 2530985 NaNs (78.2%).\n",
"\t\t\t 657514 zeros (20.3%).\n",
"Column FHINS4C has 2608417 NaNs (80.6%).\n",
"\t\t\t 576508 zeros (17.8%).\n",
"Column FHINS5C has 3129231 NaNs (96.7%).\n",
"\t\t\t 106093 zeros (3.3%).\n",
"cols_with_nans_and_zeros=['INTP', 'MLPA', 'MLPB', 'MLPCD', 'MLPE', 'MLPFG', 'MLPH', 'MLPI', 'MLPJ', 'MLPK', 'OIP', 'PAP', 'RETP', 'SEMP', 'SSIP', 'SSP', 'WAGP', 'OC', 'PERNP', 'PINCP', 'POVPIP', 'RC', 'FHINS3C', 'FHINS4C', 'FHINS5C']\n"
]
}
],
"source": [
"cols_with_nans = []\n",
"cols_with_nans_and_zeros = []\n",
"\n",
"for col in acs_data_numeric.columns:\n",
" col_data = acs_data_numeric[col]\n",
" n_nans = np.sum(np.isnan(col_data))\n",
" if n_nans > 0:\n",
" cols_with_nans.append(col)\n",
" print(f\"Column {col:10} has {n_nans:7} NaNs ({n_nans / len(col_data):.1%}).\")\n",
"\n",
" # Print number of zeros (to see overlap with representing NaNs with 0s)\n",
" n_zeros = np.sum(np.isclose(col_data, 0))\n",
" if n_zeros > 0:\n",
" print(f\"\\t\\t\\t {n_zeros} zeros ({n_zeros / len(col_data):.1%}).\")\n",
" cols_with_nans_and_zeros.append(col)\n",
"\n",
"print(f\"{cols_with_nans_and_zeros=}\")"
]
},
{
"cell_type": "markdown",
"id": "97e365f9-2878-42b0-9b5b-3a1a74b4c874",
"metadata": {},
"source": [
"## Check which tasks are compromised\n",
"> The NaNs may be on data unused by the task..."
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "362dd290-c605-4b7e-a41f-641fc2b88837",
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"{'ACSIncome': <folktables.folktables.BasicProblem at 0x11cd8b6d0>,\n",
" 'ACSEmployment': <folktables.folktables.BasicProblem at 0x10ac7b510>,\n",
" 'ACSHealthInsurance': <folktables.folktables.BasicProblem at 0x11c6d2990>,\n",
" 'ACSPublicCoverage': <folktables.folktables.BasicProblem at 0x11cd8ba50>,\n",
" 'ACSTravelTime': <folktables.folktables.BasicProblem at 0x11cd8bb50>,\n",
" 'ACSMobility': <folktables.folktables.BasicProblem at 0x11cd8bb10>,\n",
" 'ACSEmploymentFiltered': <folktables.folktables.BasicProblem at 0x11cd8bd50>,\n",
" 'ACSIncomePovertyRatio': <folktables.folktables.BasicProblem at 0x10bd244d0>}"
]
},
"execution_count": 8,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"import folktables\n",
"from folktables import BasicProblem\n",
"\n",
"acs_tasks = {name: obj for name, obj in folktables.__dict__.items() if isinstance(obj, BasicProblem)}\n",
"acs_tasks"
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "48c030a3-3a15-42d3-9f38-e4f2bd412d2c",
"metadata": {},
"outputs": [
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "cb51e9f390154f2982f03db3b29010f8",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
" 0%| | 0/8 [00:00<?, ?it/s]"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"ACSEmployment .MIG :: NaNs==0.9%\n",
"ACSEmployment .SCHL :: NaNs==2.9%\n",
"ACSEmployment .MIL :: NaNs==19.0%\n",
"ACSEmployment .DREM :: NaNs==5.0%\n",
"ACSEmployment .ESR :: NaNs==17.8%\n",
"ACSEmployment .ESP :: NaNs==80.9%\n",
"ACSHealthInsurance .MIG :: NaNs==0.9%\n",
"ACSHealthInsurance .MIL :: NaNs==19.0%\n",
"ACSHealthInsurance .PINCP :: NaNs==16.5% :: 0s==10.4% -- compromised\n",
"ACSHealthInsurance .SCHL :: NaNs==2.9%\n",
"ACSHealthInsurance .DREM :: NaNs==5.0%\n",
"ACSHealthInsurance .ESR :: NaNs==17.8%\n",
"ACSHealthInsurance .FER :: NaNs==78.4%\n",
"ACSHealthInsurance .ESP :: NaNs==80.9%\n",
"ACSPublicCoverage .MIL :: NaNs==7.0%\n",
"ACSPublicCoverage .ESR :: NaNs==3.5%\n",
"ACSPublicCoverage .FER :: NaNs==59.7%\n",
"ACSPublicCoverage .ESP :: NaNs==90.5%\n",
"ACSTravelTime .JWMNP :: NaNs==5.7%\n",
"ACSTravelTime .POVPIP :: NaNs==1.0% :: 0s==0.0% -- compromised\n",
"ACSTravelTime .ESP :: NaNs==99.4%\n",
"ACSMobility .WKHP :: NaNs==17.9%\n",
"ACSMobility .COW :: NaNs==10.9%\n",
"ACSMobility .JWMNP :: NaNs==29.8%\n",
"ACSMobility .GCL :: NaNs==67.2%\n",
"ACSMobility .ESP :: NaNs==100.0%\n",
"ACSEmploymentFiltered.GCL :: NaNs==19.8%\n",
"ACSEmploymentFiltered.ESP :: NaNs==98.6%\n",
"ACSIncomePovertyRatio.WKHP :: NaNs==48.1%\n",
"ACSIncomePovertyRatio.MIG :: NaNs==0.9%\n",
"ACSIncomePovertyRatio.MIL :: NaNs==19.0%\n",
"ACSIncomePovertyRatio.SCHL :: NaNs==2.9%\n",
"ACSIncomePovertyRatio.DREM :: NaNs==5.0%\n",
"ACSIncomePovertyRatio.GCL :: NaNs==34.8%\n",
"ACSIncomePovertyRatio.ESR :: NaNs==17.8%\n",
"ACSIncomePovertyRatio.POVPIP :: NaNs==4.2% :: 0s==1.8% -- compromised\n",
"ACSIncomePovertyRatio.OCCP :: NaNs==40.9%\n",
"ACSIncomePovertyRatio.ESP :: NaNs==80.9%\n"
]
}
],
"source": [
"from tqdm.auto import tqdm\n",
"compromised_cols_per_task = {}\n",
"\n",
"for task_name, task in tqdm(acs_tasks.items()):\n",
" # Filter out data not used for this task\n",
" task_data = task._preprocess(acs_data)\n",
"\n",
" # Check if task_data has any NaNs and 0s overlap\n",
" compromised_cols = []\n",
" for col in (set(task_data.columns) & (set(task.features) | {task.target})):\n",
" if col not in numeric_cols:\n",
" continue\n",
"\n",
" col_data = task_data[col]\n",
" n_nans = np.sum(np.isnan(col_data))\n",
" n_zeros = np.sum(np.isclose(col_data, 0))\n",
" if n_nans > 0:\n",
" print(f\"{task_name:20}.{col:5} :: NaNs=={n_nans / len(col_data):.1%}\", end=\"\")\n",
" if n_zeros > 0:\n",
" print(f\" :: 0s=={n_zeros / len(col_data):.1%} -- compromised\", end=\"\")\n",
" compromised_cols.append(col)\n",
"\n",
" print()\n",
" \n",
"\n",
" compromised_cols_per_task[task_name] = compromised_cols\n"
]
},
{
"cell_type": "markdown",
"id": "5cfca50e-22d3-446e-aa16-f9be60fca0ea",
"metadata": {},
"source": [
"---"
]
}
],
"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": "3.11.9"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment