breast-cancer-raps-grid-search.ipynb

{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.14","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"none","dataSources":[{"sourceId":9245678,"sourceType":"datasetVersion","datasetId":5593089},{"sourceId":9349083,"sourceType":"datasetVersion","datasetId":5666853},{"sourceId":9349741,"sourceType":"datasetVersion","datasetId":5663356},{"sourceId":9349860,"sourceType":"datasetVersion","datasetId":5667477},{"sourceId":9350027,"sourceType":"datasetVersion","datasetId":5667599},{"sourceId":9350189,"sourceType":"datasetVersion","datasetId":5667722}],"dockerImageVersionId":30761,"isInternetEnabled":true,"language":"python","sourceType":"notebook","isGpuEnabled":false}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# Importing necessary libraries","metadata":{}},{"cell_type":"code","source":"import pandas as pd\nfrom sklearn.model_selection import train_test_split\nimport torch\nimport numpy as np","metadata":{"execution":{"iopub.status.busy":"2024-09-09T11:11:29.282783Z","iopub.execute_input":"2024-09-09T11:11:29.283230Z","iopub.status.idle":"2024-09-09T11:11:29.289571Z","shell.execute_reply.started":"2024-09-09T11:11:29.283190Z","shell.execute_reply":"2024-09-09T11:11:29.288125Z"},"trusted":true},"execution_count":8,"outputs":[]},{"cell_type":"markdown","source":"# CP method metrices(specific to Breast Cancer Dataset):- ","metadata":{}},{"cell_type":"code","source":"def avg_set_size_metric(conformal_set):\n    lengths = torch.sum(conformal_set, dim=1)\n    avg_set_size_len = torch.sum(lengths)/conformal_set.shape[0]\n    return avg_set_size_len\n\n\n\n\ndef breast_cancer_class_Overlap_metric(conformal_set, label):\n\n    overlap_count = 0\n\n    for i in range(conformal_set.shape[0]):\n        current_set = conformal_set[i]\n\n        if (label[i] == 0 or label[i] == 1 or label[i] == 2 or label[i] == 3 ) and (current_set[4] == 1 or current_set[5] == 1 or current_set[6] == 1 or current_set[7] == 1):\n            overlap_count += 1\n        \n        elif (label[i] == 4 or label[i] == 5 or label[i] == 6 or label[i] == 7 ) and (current_set[0] == 1 or current_set[1] == 1 or current_set[2] == 1 or current_set[3] == 1):\n            overlap_count += 1\n\n\n    perecentage_of_overlap  = (overlap_count/conformal_set.shape[0])*100\n\n    return perecentage_of_overlap\n\n\n\n\ndef breast_cancer_confusion_set_Overlap_metric(conformal_set, label):\n\n    overlap_count = 0\n\n    for i in range(conformal_set.shape[0]):\n        current_set = conformal_set[i]\n\n        if (label[i] == 1) and (current_set[2] == 1):\n            overlap_count += 1\n\n        elif (label[i] == 2) and (current_set[1] == 1):\n            overlap_count += 1\n\n        \n\n        elif (label[i] == 4) and (current_set[5] == 1):\n            overlap_count += 1\n\n        elif (label[i] == 5) and (current_set[4] == 1):\n            overlap_count += 1\n            \n            \n    perecentage_of_overlap  = (overlap_count/conformal_set.shape[0])*100\n\n    return perecentage_of_overlap","metadata":{"execution":{"iopub.status.busy":"2024-09-09T12:53:45.181801Z","iopub.execute_input":"2024-09-09T12:53:45.182287Z","iopub.status.idle":"2024-09-09T12:53:45.198658Z","shell.execute_reply.started":"2024-09-09T12:53:45.182228Z","shell.execute_reply":"2024-09-09T12:53:45.197333Z"},"trusted":true},"execution_count":56,"outputs":[]},{"cell_type":"markdown","source":"# RAPS class :- ","metadata":{}},{"cell_type":"code","source":"class RAPS():\n    def __init__(self, softmax, true_class, alpha, k_reg, lambd, rand=True):\n        self.prob_output = softmax\n        self.true_class = true_class\n        self.alpha = (1 - alpha) * (1 + (1 / softmax.shape[0]))\n        self.k_reg = k_reg\n        self.lambd = lambd\n        self.rand = rand\n        \n        \n        \n    def conformal_score(self):\n        conformal_score = []\n        for i in range(self.prob_output.shape[0]):\n            true_class_prob = self.prob_output[i][self.true_class[i]]\n            current_class_prob = self.prob_output[i]\n            sorted_class_prob, _ = torch.sort(current_class_prob, descending=True)\n            index = torch.nonzero(sorted_class_prob == true_class_prob).item()\n            cumulative_sum = torch.sum(sorted_class_prob[:index + 1])\n            \n            if index - self.k_reg > 0:\n               cumulative_sum = cumulative_sum + self.lambd*(index - self.k_reg)\n            \n            if self.rand:\n                U = torch.rand(1).item()\n                cumulative_sum = cumulative_sum - U*sorted_class_prob[index]\n\n            conformal_score.append(cumulative_sum)\n        \n        conformal_score = torch.tensor(conformal_score)\n        \n        return conformal_score\n    \n    \n    \n    def quantile(self):\n        conformal_scores = self.conformal_score()\n        quantile_value = torch.quantile(conformal_scores, self.alpha)\n        return quantile_value\n    \n    \n    \n    def prediction(self, softmax, quantile_value):\n        prob_output = softmax\n        prediction = torch.zeros(prob_output.shape[0], prob_output.shape[1])\n        for i in range(prob_output.shape[0]):\n            current_class_prob = prob_output[i]\n            sorted_class_prob, _ = torch.sort(current_class_prob, descending=True)\n            sum = 0\n            j = 0\n            for idx in range(len(sorted_class_prob)):\n                if sum <= quantile_value:\n                    sum += sorted_class_prob[idx]\n                    if idx - self.k_reg > 0:\n                        sum = sum + self.lambd*(idx - self.k_reg)\n                    j += 1\n                else:\n                    break\n                    \n            \"\"\"\n            \n            if self.rand:\n                U = torch.rand(1).item()\n                if j != prob_output.shape[1]:\n                    N = torch.sum(sorted_class_prob[:j + 1]) - quantile_value\n                else:\n                    N = torch.sum(sorted_class_prob[:j]) - quantile_value\n                if idx - self.k_reg > 0:\n                    N += self.lambd*(j - self.k_reg)\n\n                if j != prob_output.shape[1]:\n                    D = sorted_class_prob[j]\n                else:\n                    D = sorted_class_prob[j-1]\n                if idx - self.k_reg > 0:\n                    D += self.lambd\n\n                if N/D <= U:\n                    j = j -1\n                    \n                \"\"\" \n        \n            \n            \n            \n                    \n            for idx in range(j):\n                index = torch.nonzero(current_class_prob == sorted_class_prob[idx]).item()\n                prediction[i][index] = 1.0\n                \n        return prediction","metadata":{"execution":{"iopub.status.busy":"2024-09-09T12:53:47.008892Z","iopub.execute_input":"2024-09-09T12:53:47.009378Z","iopub.status.idle":"2024-09-09T12:53:47.029299Z","shell.execute_reply.started":"2024-09-09T12:53:47.009334Z","shell.execute_reply":"2024-09-09T12:53:47.027897Z"},"trusted":true},"execution_count":57,"outputs":[]},{"cell_type":"markdown","source":"# Main function :- ","metadata":{}},{"cell_type":"code","source":"def main(expt_no, Trials, alpha,k_reg, lambd, rand):\n    \n    avg_set_size_len_for_T_trials = []\n    avg_coverage_gap_for_T_trials = []\n    avg_coverage_for_T_trials = []\n\n    normal_avg_set_size_len_for_T_trials = []\n    abnormal_avg_set_size_len_for_T_trials = []\n\n\n    perecentage_of_overlap_for_T_trials = []\n\n    confusion_set_Overlap_metric_for_T_trials = []\n    \n    \n    \n    for t in range(Trials):\n        \n        # loading the annotation file :-\n        #Expt1:-\n        #path = ''\n        #df = pd.read_csv(path)\n        \n        #Expt2:-\n        #path = ''\n        #df = pd.read_csv(path)\n        \n        #Expt3:-\n        #path = ''\n        #df = pd.read_csv(path)\n        \n        #Expt4:-\n        #path = ''\n        #df = pd.read_csv(path)\n        \n        \n        # calib-test split :- \n        feature_test, feature_calib = train_test_split(df, test_size = 0.3, stratify=df['Label'], random_state=42)\n\n        feature_test = feature_test.reset_index(drop=True)\n        feature_calib = feature_calib.reset_index(drop=True)\n\n        prob_output = feature_calib.iloc[:,:-1]\n        df_np = prob_output.values \n        df_prob_output_calib = torch.tensor(df_np, dtype=torch.float32)\n\n        prob_output = feature_test.iloc[:,:-1]\n        df_np = prob_output.values\n        df_prob_output_test = torch.tensor(df_np, dtype=torch.float32)\n\n\n        true_class = feature_calib.iloc[:,-1]\n        df_np = true_class.values\n        df_true_class_calib = torch.tensor(df_np, dtype=torch.int)\n\n\n        true_class = feature_test.iloc[:,-1]\n        df_np = true_class.values\n        df_true_class_test = torch.tensor(df_np, dtype=torch.int)\n        \n        conformal_wrapper = RAPS(df_prob_output_calib, df_true_class_calib, alpha, k_reg, lambd, rand)\n        quantile_value = conformal_wrapper.quantile()\n        \n        conformal_set = conformal_wrapper.prediction(df_prob_output_test, quantile_value)\n        \n        \n        \n        \n        if expt_no == 1:\n            avg_set_size = avg_set_size_metric(conformal_set)\n            #print(f'avg_set_size:- {avg_set_size}')\n            avg_set_size_len_for_T_trials.append(avg_set_size)\n        \n        \n        elif expt_no == 2:\n            label = df_true_class_test\n            indices_0 = torch.nonzero(label == 0).squeeze()\n            indices_1 = torch.nonzero(label == 1).squeeze()\n            indices_2 = torch.nonzero(label == 2).squeeze()\n            indices_3 = torch.nonzero(label == 3).squeeze()\n            indices_4 = torch.nonzero(label == 4).squeeze()\n            indices_5 = torch.nonzero(label == 5).squeeze()\n            indices_6 = torch.nonzero(label == 6).squeeze()\n            indices_7 = torch.nonzero(label == 7).squeeze()\n            \n            \n            Normal_idx = torch.cat((indices_0, indices_1, indices_2, indices_3))\n            Abnormal_idx = torch.cat((indices_4, indices_5, indices_6, indices_7))\n\n            normal_conformal_prediction_set = conformal_set[Normal_idx, :]\n            abnormal_conformal_prediction_set = conformal_set[Abnormal_idx, :]\n\n            normal_avg_set_size_len = avg_set_size_metric(normal_conformal_prediction_set)\n            abnormal_avg_set_size_len = avg_set_size_metric(abnormal_conformal_prediction_set)\n\n            normal_avg_set_size_len_for_T_trials.append(normal_avg_set_size_len)\n            abnormal_avg_set_size_len_for_T_trials.append(abnormal_avg_set_size_len)\n            \n            \n        \n        elif expt_no == 3:\n            perecentage_of_overlap = breast_cancer_class_Overlap_metric(conformal_set, df_true_class_test)\n            \n            perecentage_of_overlap_for_T_trials.append(perecentage_of_overlap)\n\n\n        elif expt_no == 4:\n            perecentage_of_confusion = breast_cancer_confusion_set_Overlap_metric(conformal_set, df_true_class_test)\n            #print(f'perecentage_of_confusion :- {perecentage_of_confusion}')\n\n            confusion_set_Overlap_metric_for_T_trials.append(perecentage_of_confusion)\n            \n            \n            \n    \n    \n    \n    \n    if expt_no == 1:\n        avg_set_size_len_for_T_trials = np.array(avg_set_size_len_for_T_trials)\n        average = np.mean(avg_set_size_len_for_T_trials)\n        std_dev = np.std(avg_set_size_len_for_T_trials, ddof=1)\n\n        \n        #print(f\"Average set_size_len_for_T_trials: {average}\")\n        #print(f\"Standard Deviation set_size_len_for_T_trials: {std_dev}\")\n        \n        return average, std_dev\n        \n        \n    \n    \n    elif expt_no == 2:\n        \n\n        normal_avg_set_size_len_for_T_trials = np.array(normal_avg_set_size_len_for_T_trials)\n        normal_average_set_size_len = np.mean(normal_avg_set_size_len_for_T_trials)\n        normal_std_dev_set_size_len = np.std(normal_avg_set_size_len_for_T_trials, ddof=1)\n\n        \n\n\n\n        abnormal_avg_set_size_len_for_T_trials = np.array(abnormal_avg_set_size_len_for_T_trials)\n        abnormal_average_set_size_len = np.mean(abnormal_avg_set_size_len_for_T_trials)\n        abnormal_std_dev_set_size_len = np.std(abnormal_avg_set_size_len_for_T_trials, ddof=1)\n\n        \n        \n        return normal_average_set_size_len, normal_std_dev_set_size_len, abnormal_average_set_size_len, abnormal_std_dev_set_size_len\n\n        \n        \n        \n\n    elif expt_no == 3:\n        perecentage_of_overlap_for_T_trials = np.array(perecentage_of_overlap_for_T_trials)\n        average_perecentage_of_overlap_for_T_trials = np.mean(perecentage_of_overlap_for_T_trials)\n        std_dev_perecentage_of_overlap_for_T_trials = np.std(perecentage_of_overlap_for_T_trials, ddof=1)\n\n       \n        \n        return average_perecentage_of_overlap_for_T_trials, std_dev_perecentage_of_overlap_for_T_trials\n\n        \n        \n        \n        \n\n    elif expt_no == 4:\n        confusion_set_Overlap_metric_for_T_trials = np.array(confusion_set_Overlap_metric_for_T_trials)\n        average_confusion_set_Overlap_metric_for_T_trials = np.mean(confusion_set_Overlap_metric_for_T_trials)\n        std_dev_confusion_set_Overlap_metric_for_T_trials = np.std(confusion_set_Overlap_metric_for_T_trials, ddof=1)\n\n     \n        \n        \n        return average_confusion_set_Overlap_metric_for_T_trials, std_dev_confusion_set_Overlap_metric_for_T_trials\n        ","metadata":{"execution":{"iopub.status.busy":"2024-09-09T12:53:48.463194Z","iopub.execute_input":"2024-09-09T12:53:48.464350Z","iopub.status.idle":"2024-09-09T12:53:48.494035Z","shell.execute_reply.started":"2024-09-09T12:53:48.464297Z","shell.execute_reply":"2024-09-09T12:53:48.492726Z"},"trusted":true},"execution_count":58,"outputs":[]},{"cell_type":"markdown","source":"# Grid search for hyperparameter tuning(k_reg, lambd) for RAPS method","metadata":{}},{"cell_type":"code","source":"expt_no = 4\nTrials = 10\nalpha = 0.05\nk_reg = [1, 2, 3, 4, 5, 6, 7]\nlambd = [0, 0.0001, 0.001, 0.01, 0.02, 0.05, 0.2, 0.5, 0.7, 1.0]\nrand = True\n\n\n\ndic_expt_1 = {\n    'average_len' : [],\n    'std_dev_len' : [],\n }\n\n\ndic_expt_3 = {\n    'average_perecentage_of_overlap_for_T_trials' : [],\n    'std_dev_perecentage_of_overlap_for_T_trials' : []\n}\n\n\ndic_expt_4 = {\n    'average_confusion_set_Overlap_metric_for_T_trials' : [],\n    'std_dev_confusion_set_Overlap_metric_for_T_trials' : []\n   \n}\n\n\n\ndic_expt_2 = {\n    'normal_average_set_size_len' : [],\n    'normal_std_dev_set_size_len' : [],\n    'abnormal_average_set_size_len' : [],\n    'abnormal_std_dev_set_size_len' : []\n}\n    \n    \n    \n    \nfor i in range(len(k_reg)):\n    for j in range(len(lambd)):\n        print(f'k_reg :- {k_reg[i]}, lambd :- {lambd[j]}')\n        \n        main(expt_no, Trials, alpha, k_reg[i], lambd[j], rand)\n        \n        if expt_no == 1:\n            average_len, std_dev_len = main(expt_no, Trials, alpha, k_reg[i], lambd[j], rand)\n            dic_expt_1['average_len'].append(average_len)\n            dic_expt_1['std_dev_len'].append(std_dev_len)\n            \n        elif expt_no == 2:\n            normal_average_set_size_len, normal_std_dev_set_size_len, abnormal_average_set_size_len, abnormal_std_dev_set_size_len = main(expt_no, Trials, alpha, k_reg[i], lambd[j], rand)\n            dic_expt_2['normal_average_set_size_len'].append(normal_average_set_size_len)\n            dic_expt_2['normal_std_dev_set_size_len'].append(normal_std_dev_set_size_len)\n            dic_expt_2['abnormal_average_set_size_len'].append(abnormal_average_set_size_len)\n            dic_expt_2['abnormal_std_dev_set_size_len'].append(abnormal_std_dev_set_size_len)\n            \n            \n        elif expt_no == 3:\n            average_perecentage_of_overlap_for_T_trials, std_dev_perecentage_of_overlap_for_T_trials = main(expt_no, Trials, alpha, k_reg[i], lambd[j], rand) \n            dic_expt_3['average_perecentage_of_overlap_for_T_trials'].append(average_perecentage_of_overlap_for_T_trials)\n            dic_expt_3['std_dev_perecentage_of_overlap_for_T_trials'].append(std_dev_perecentage_of_overlap_for_T_trials)\n            \n            \n        elif expt_no == 4:\n            average_confusion_set_Overlap_metric_for_T_trials, std_dev_confusion_set_Overlap_metric_for_T_trials = main(expt_no, Trials, alpha, k_reg[i], lambd[j], rand)\n            dic_expt_4['average_confusion_set_Overlap_metric_for_T_trials'].append(average_confusion_set_Overlap_metric_for_T_trials)\n            dic_expt_4['std_dev_confusion_set_Overlap_metric_for_T_trials'].append(std_dev_confusion_set_Overlap_metric_for_T_trials)\n            \n        ","metadata":{"execution":{"iopub.status.busy":"2024-09-09T13:07:06.089256Z","iopub.execute_input":"2024-09-09T13:07:06.090311Z","iopub.status.idle":"2024-09-09T13:14:34.836539Z","shell.execute_reply.started":"2024-09-09T13:07:06.090260Z","shell.execute_reply":"2024-09-09T13:14:34.835302Z"},"trusted":true},"execution_count":62,"outputs":[{"name":"stdout","text":"k_reg :- 1, lambd :- 0\nk_reg :- 1, lambd :- 0.0001\nk_reg :- 1, lambd :- 0.001\nk_reg :- 1, lambd :- 0.01\nk_reg :- 1, lambd :- 0.02\nk_reg :- 1, lambd :- 0.05\nk_reg :- 1, lambd :- 0.2\nk_reg :- 1, lambd :- 0.5\nk_reg :- 1, lambd :- 0.7\nk_reg :- 1, lambd :- 1.0\nk_reg :- 2, lambd :- 0\nk_reg :- 2, lambd :- 0.0001\nk_reg :- 2, lambd :- 0.001\nk_reg :- 2, lambd :- 0.01\nk_reg :- 2, lambd :- 0.02\nk_reg :- 2, lambd :- 0.05\nk_reg :- 2, lambd :- 0.2\nk_reg :- 2, lambd :- 0.5\nk_reg :- 2, lambd :- 0.7\nk_reg :- 2, lambd :- 1.0\nk_reg :- 3, lambd :- 0\nk_reg :- 3, lambd :- 0.0001\nk_reg :- 3, lambd :- 0.001\nk_reg :- 3, lambd :- 0.01\nk_reg :- 3, lambd :- 0.02\nk_reg :- 3, lambd :- 0.05\nk_reg :- 3, lambd :- 0.2\nk_reg :- 3, lambd :- 0.5\nk_reg :- 3, lambd :- 0.7\nk_reg :- 3, lambd :- 1.0\nk_reg :- 4, lambd :- 0\nk_reg :- 4, lambd :- 0.0001\nk_reg :- 4, lambd :- 0.001\nk_reg :- 4, lambd :- 0.01\nk_reg :- 4, lambd :- 0.02\nk_reg :- 4, lambd :- 0.05\nk_reg :- 4, lambd :- 0.2\nk_reg :- 4, lambd :- 0.5\nk_reg :- 4, lambd :- 0.7\nk_reg :- 4, lambd :- 1.0\nk_reg :- 5, lambd :- 0\nk_reg :- 5, lambd :- 0.0001\nk_reg :- 5, lambd :- 0.001\nk_reg :- 5, lambd :- 0.01\nk_reg :- 5, lambd :- 0.02\nk_reg :- 5, lambd :- 0.05\nk_reg :- 5, lambd :- 0.2\nk_reg :- 5, lambd :- 0.5\nk_reg :- 5, lambd :- 0.7\nk_reg :- 5, lambd :- 1.0\nk_reg :- 6, lambd :- 0\nk_reg :- 6, lambd :- 0.0001\nk_reg :- 6, lambd :- 0.001\nk_reg :- 6, lambd :- 0.01\nk_reg :- 6, lambd :- 0.02\nk_reg :- 6, lambd :- 0.05\nk_reg :- 6, lambd :- 0.2\nk_reg :- 6, lambd :- 0.5\nk_reg :- 6, lambd :- 0.7\nk_reg :- 6, lambd :- 1.0\nk_reg :- 7, lambd :- 0\nk_reg :- 7, lambd :- 0.0001\nk_reg :- 7, lambd :- 0.001\nk_reg :- 7, lambd :- 0.01\nk_reg :- 7, lambd :- 0.02\nk_reg :- 7, lambd :- 0.05\nk_reg :- 7, lambd :- 0.2\nk_reg :- 7, lambd :- 0.5\nk_reg :- 7, lambd :- 0.7\nk_reg :- 7, lambd :- 1.0\n","output_type":"stream"}]},{"cell_type":"code","source":"df = pd.DataFrame(dic_expt_4)\ndf","metadata":{"execution":{"iopub.status.busy":"2024-09-09T13:15:28.897032Z","iopub.execute_input":"2024-09-09T13:15:28.897528Z","iopub.status.idle":"2024-09-09T13:15:28.912998Z","shell.execute_reply.started":"2024-09-09T13:15:28.897483Z","shell.execute_reply":"2024-09-09T13:15:28.911357Z"},"trusted":true},"execution_count":63,"outputs":[{"execution_count":63,"output_type":"execute_result","data":{"text/plain":"    average_confusion_set_Overlap_metric_for_T_trials  \\\n0                                           23.142685   \n1                                           22.071042   \n2                                           22.221553   \n3                                           23.251054   \n4                                           24.388922   \n..                                                ...   \n65                                          22.600843   \n66                                          22.835641   \n67                                          23.594220   \n68                                          22.552679   \n69                                          22.438290   \n\n    std_dev_confusion_set_Overlap_metric_for_T_trials  \n0                                            1.592158  \n1                                            1.795899  \n2                                            1.725506  \n3                                            1.821859  \n4                                            1.749151  \n..                                                ...  \n65                                           1.829228  \n66                                           1.729143  \n67                                           2.247185  \n68                                           1.502627  \n69                                           1.759619  \n\n[70 rows x 2 columns]","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>average_confusion_set_Overlap_metric_for_T_trials</th>\n      <th>std_dev_confusion_set_Overlap_metric_for_T_trials</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>23.142685</td>\n      <td>1.592158</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>22.071042</td>\n      <td>1.795899</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>22.221553</td>\n      <td>1.725506</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>23.251054</td>\n      <td>1.821859</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>24.388922</td>\n      <td>1.749151</td>\n    </tr>\n    <tr>\n      <th>...</th>\n      <td>...</td>\n      <td>...</td>\n    </tr>\n    <tr>\n      <th>65</th>\n      <td>22.600843</td>\n      <td>1.829228</td>\n    </tr>\n    <tr>\n      <th>66</th>\n      <td>22.835641</td>\n      <td>1.729143</td>\n    </tr>\n    <tr>\n      <th>67</th>\n      <td>23.594220</td>\n      <td>2.247185</td>\n    </tr>\n    <tr>\n      <th>68</th>\n      <td>22.552679</td>\n      <td>1.502627</td>\n    </tr>\n    <tr>\n      <th>69</th>\n      <td>22.438290</td>\n      <td>1.759619</td>\n    </tr>\n  </tbody>\n</table>\n<p>70 rows × 2 columns</p>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"min_index = df['average_confusion_set_Overlap_metric_for_T_trials'].idxmin()\nmin_row = df.loc[min_index]\nmin_row","metadata":{"execution":{"iopub.status.busy":"2024-09-09T13:15:30.984183Z","iopub.execute_input":"2024-09-09T13:15:30.985354Z","iopub.status.idle":"2024-09-09T13:15:30.994344Z","shell.execute_reply.started":"2024-09-09T13:15:30.985307Z","shell.execute_reply":"2024-09-09T13:15:30.993066Z"},"trusted":true},"execution_count":64,"outputs":[{"execution_count":64,"output_type":"execute_result","data":{"text/plain":"average_confusion_set_Overlap_metric_for_T_trials    21.366647\nstd_dev_confusion_set_Overlap_metric_for_T_trials     0.875421\nName: 39, dtype: float64"},"metadata":{}}]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}