switch .ipynb files to .py & add new function res_df into package

multi_participants 2/case_3/case_3_accuracy.py

@@ -0,0 +1,65 @@
+import os
+from mind_reading_package import mind_reading as mr
+import pandas as pd
+
+# list all folders' name
+participants = os.listdir('path')
+
+# remove the 'cha' folder we don't need
+participants = participants.remove('cha')
+
+# create the initial dataframe
+df = pd.DataFrame(index = ['SVC', 'DTC', 'NB', 'NN'])
+
+for participant in participants:
+    # iterate all the folders
+
+    for file in os.listdir(participant):
+        # iterate all files in every folder, find out the one end with 'Cong.csv' and 'Incong.csv' as input data
+
+        if file.endswith('Cong.csv'): file1 = f"{participant}/{file}"
+        if file.endswith('Incong.csv'): file2 = f"{participant}/{file}"
+
+    # load in cong and incong data for them
+    df1 = mr.load_data(file1)
+    df2 = mr.load_data(file2)
+
+    # concatenate such data
+    data = mr.concatenate_data(df1, df2)
+
+    # find trials to later separate
+    trials_index = mr.find_trials(data)
+
+    # separate trials
+    trials = mr.separate_trials(data, trials_index)
+
+    # create the label column
+    labels = mr.create_binary_labels(data)
+
+    # Go through each trial, reset the columns, we split from 100-300ms ((308th sample to 513th sample))
+    pro_trials = mr.process_trials(trials)
+
+    # Find the mean across channels
+    avg_trials = mr.average_trials(pro_trials)
+
+    # concatenates the average trials dataframe with labels
+    ml_df = mr.create_ml_df(avg_trials, labels)
+
+    # train models
+    X_train, X_test, y_train, y_test = mr.prepare_ml_df(ml_df)
+
+    acc_svc, precision_svc = mr.train_svc(X_train, X_test, y_train, y_test)
+
+    acc_dtc, precision_dtc = mr.train_dtc(X_train, X_test, y_train, y_test)
+
+    acc_nb, precision_nb = mr.train_nb(X_train, X_test, y_train, y_test)
+
+    acc_nn, precision_nn = mr.train_nn(64, X_train, X_test, y_train, y_test)
+
+    # add every participant's accuracy together
+    acc_list = [f"{acc_svc:.2f}", f"{acc_dtc:.2f}", f"{acc_nb:.2f}", f"{acc_nn:.2f}"]
+
+    df = mr.res_df(df, acc_list, participant)
+
+# generate result .csv file
+df.to_csv('case_3_accuracy.csv') 

multi_participants 2/case_3/case_3_precision.py

@@ -0,0 +1,66 @@
+import os
+from mind_reading_package import mind_reading as mr
+import pandas as pd
+
+# list all folders' name
+participants = os.listdir('path')
+
+# remove the 'cha' folder we don't need
+participants = participants.remove('cha')
+
+# create the initial dataframe
+df = pd.DataFrame(index = ['SVC', 'DTC', 'NB', 'NN'])
+
+for participant in participants:
+    # iterate all the folders
+
+    for file in os.listdir(participant):
+        # iterate all files in every folder, find out the one end with 'Cong.csv' and 'Incong.csv' as input data
+
+        if file.endswith('Cong.csv'): file1 = f"{participant}/{file}"
+        if file.endswith('Incong.csv'): file2 = f"{participant}/{file}"
+
+    # load in cong and incong data for them
+    df1 = mr.load_data(file1)
+    df2 = mr.load_data(file2)
+
+    # concatenate such data
+    data = mr.concatenate_data(df1, df2)
+
+    # find trials to later separate
+    trials_index = mr.find_trials(data)
+
+    # separate trials
+    trials = mr.separate_trials(data, trials_index)
+
+    # create the label column
+    labels = mr.create_binary_labels(data)
+
+    # Go through each trial, reset the columns, we split from 100-300ms ((308th sample to 513th sample))
+    pro_trials = mr.process_trials(trials)
+
+    # Find the mean across channels
+    avg_trials = mr.average_trials(pro_trials)
+
+    # concatenates the average trials dataframe with labels
+    ml_df = mr.create_ml_df(avg_trials, labels)
+
+    # train models
+    X_train, X_test, y_train, y_test = mr.prepare_ml_df(ml_df)
+
+    acc_svc, precision_svc = mr.train_svc(X_train, X_test, y_train, y_test)
+
+    acc_dtc, precision_dtc = mr.train_dtc(X_train, X_test, y_train, y_test)
+
+    acc_nb, precision_nb = mr.train_nb(X_train, X_test, y_train, y_test)
+
+    acc_nn, precision_nn = mr.train_nn(64, X_train, X_test, y_train, y_test)
+
+    # add every participant's precision together
+    precision_list = [f"{precision_svc:.2f}", f"{precision_dtc:.2f}", f"{precision_nb:.2f}", f"{precision_nn:.2f}"]
+
+    df = mr.res_df(df, precision_list, participant)
+
+# generate result .csv file
+df.to_csv('case_3_accuracy.csv')
+

multi_participants 2/case_4/case_4_accuracy.py

@@ -0,0 +1,65 @@
+import os
+from mind_reading_package import mind_reading as mr
+import pandas as pd
+
+# list all folders' name
+participants = os.listdir('path')
+
+# remove the 'cha' folder we don't need
+participants = participants.remove('cha')
+
+# create the initial dataframe
+df = pd.DataFrame(index = ['SVC', 'DTC', 'NB', 'NN'])
+
+for participant in participants:
+    # iterate all the folders
+
+    for file in os.listdir(participant):
+        # iterate all files in every folder, find out the one end with 'Cong.csv' and 'Incong.csv' as input data
+
+        if file.endswith('Cong.csv'): file1 = f"{participant}/{file}"
+        if file.endswith('Incong.csv'): file2 = f"{participant}/{file}"
+
+    # load in cong and incong data for them
+    df1 = mr.load_data(file1)
+    df2 = mr.load_data(file2)
+
+    # concatenate such data
+    data = mr.concatenate_data(df1, df2)
+
+    # find trials to later separate
+    trials_index = mr.find_trials(data)
+
+    # separate trials
+    trials = mr.separate_trials(data, trials_index)
+
+    # create the label column
+    labels = mr.create_multi_labels(data)
+
+    # Go through each trial, reset the columns, we split from 100-300ms ((308th sample to 513th sample))
+    pro_trials = mr.process_trials(trials)
+
+    # Find the mean across channels
+    avg_trials = mr.average_trials(pro_trials)
+
+    # concatenates the average trials dataframe with labels
+    ml_df = mr.create_ml_df(avg_trials, labels)
+
+    # train models
+    X_train, X_test, y_train, y_test = mr.prepare_ml_df(ml_df)
+
+    acc_svc, precision_svc = mr.train_svc_multi(X_train, X_test, y_train, y_test)
+
+    acc_dtc, precision_dtc = mr.train_dtc_multi(X_train, X_test, y_train, y_test)
+
+    acc_nb, precision_nb = mr.train_nb_multi(X_train, X_test, y_train, y_test)
+
+    acc_nn, precision_nn = mr.train_nn_multi(64, X_train, X_test, y_train, y_test)
+
+    # add every participant's accuracy together
+    acc_list = [f"{acc_svc:.2f}", f"{acc_dtc:.2f}", f"{acc_nb:.2f}", f"{acc_nn:.2f}"]
+
+    df = mr.res_df(df, acc_list, participant)
+
+# generate result .csv file
+df.to_csv('case_4_accuracy.csv')

multi_participants 2/case_4/case_4_precision.py

@@ -0,0 +1,65 @@
+import os
+from mind_reading_package import mind_reading as mr
+import pandas as pd
+
+# list all folders' name
+participants = os.listdir('path')
+
+# remove the 'cha' folder we don't need
+participants = participants.remove('cha')
+
+# create the initial dataframe
+df = pd.DataFrame(index = ['SVC', 'DTC', 'NB', 'NN'])
+
+for participant in participants:
+    # iterate all the folders
+
+    for file in os.listdir(participant):
+        # iterate all files in every folder, find out the one end with 'Cong.csv' and 'Incong.csv' as input data
+
+        if file.endswith('Cong.csv'): file1 = f"{participant}/{file}"
+        if file.endswith('Incong.csv'): file2 = f"{participant}/{file}"
+
+    # load in cong and incong data for them
+    df1 = mr.load_data(file1)
+    df2 = mr.load_data(file2)
+
+    # concatenate such data
+    data = mr.concatenate_data(df1, df2)
+
+    # find trials to later separate
+    trials_index = mr.find_trials(data)
+
+    # separate trials
+    trials = mr.separate_trials(data, trials_index)
+
+    # create the label column
+    labels = mr.create_multi_labels(data)
+
+    # Go through each trial, reset the columns, we split from 100-300ms ((308th sample to 513th sample))
+    pro_trials = mr.process_trials(trials)
+
+    # Find the mean across channels
+    avg_trials = mr.average_trials(pro_trials)
+
+    # concatenates the average trials dataframe with labels
+    ml_df = mr.create_ml_df(avg_trials, labels)
+
+    # train models
+    X_train, X_test, y_train, y_test = mr.prepare_ml_df(ml_df)
+
+    acc_svc, precision_svc = mr.train_svc_multi(X_train, X_test, y_train, y_test)
+
+    acc_dtc, precision_dtc = mr.train_dtc_multi(X_train, X_test, y_train, y_test)
+
+    acc_nb, precision_nb = mr.train_nb_multi(X_train, X_test, y_train, y_test)
+
+    acc_nn, precision_nn = mr.train_nn_multi(64, X_train, X_test, y_train, y_test)
+
+    # add every participant's precision together
+    precision_list = [f"{precision_svc:.2f}", f"{precision_dtc:.2f}", f"{precision_nb:.2f}", f"{precision_nn:.2f}"]
+
+    df = mr.res_df(df, precision_list, participant)
+
+# generate result .csv file
+df.to_csv('case_4_precision.csv')

multi_participants/mind-reading package/mind_reading.py → multi_participants 2/mind_reading_package/mind_reading.py

@@ -310,3 +310,18 @@ def create_metric_df(acc_list, prec_list, model_list):
     metric_df.columns = ['acc', 'prec']
 
     return metric_df
+
+def res_df(df, column, participant):
+    '''
+    Add precision/accuracy for every participant to the whole results
+    Args:
+        df: the dataframe of all results we have had
+        column: the dataframe of result we want to add
+        participant: participant number
+    returns:
+        all precision/accuracy results
+    '''
+
+    data = pd.DataFrame({f"Participant {participant}": column})
+    df[f"Participant {participant}"] = data[f"Participant {participant}"].values
+    return df

multi_participants/mind-reading package/mind_reading_v2.py → multi_participants 2/mind_reading_package/mind_reading_v2.py

@@ -512,3 +512,18 @@ def train_nn_multi(n_inputs, X_train, X_test, y_train, y_test):
     _, accuracy, precision = model.evaluate(X_test, y_test, verbose=0)
 
     return accuracy, precision
+
+def res_df(df, column, participant):
+    '''
+    Add precision/accuracy for every participant to the whole results
+    Args:
+        df: the dataframe of all results we have had
+        column: the dataframe of result we want to add
+        participant: participant number
+    returns:
+        all precision/accuracy results
+    '''
+
+    data = pd.DataFrame({f"Participant {participant}": column})
+    df[f"Participant {participant}"] = data[f"Participant {participant}"].values
+    return df