From 481361b7e0801bc695b0c7cc9c3e7d1e23627d91 Mon Sep 17 00:00:00 2001 From: Svetlnkk <89974865+Svetlnkk@users.noreply.github.com> Date: Thu, 26 Oct 2023 11:27:02 +0400 Subject: [PATCH] lal --- zavrazhnova_svetlana_lab_6/fraud_dataset.csv | 87 +++++++++++++++++++ .../zavrazhnova_svetlana_lab_6.py | 84 ++++++++++++++++++ 2 files changed, 171 insertions(+) create mode 100644 zavrazhnova_svetlana_lab_6/fraud_dataset.csv create mode 100644 zavrazhnova_svetlana_lab_6/zavrazhnova_svetlana_lab_6.py diff --git a/zavrazhnova_svetlana_lab_6/fraud_dataset.csv b/zavrazhnova_svetlana_lab_6/fraud_dataset.csv new file mode 100644 index 0000000..f23b91d --- /dev/null +++ b/zavrazhnova_svetlana_lab_6/fraud_dataset.csv @@ -0,0 +1,87 @@ +transaction_id,transaction_amount,location,merchant,age,gender,fraud_label +1,1000.00,New York,ABC Corp,35,M,0 +2,500.00,Chicago,XYZ Inc,45,F,0 +3,2000.00,Los Angeles,ABC Corp,28,M,1 +4,1500.00,San Francisco,XYZ Inc,30,F,0 +5,800.00,Chicago,ABC Corp,50,F,0 +6,3000.00,New York,XYZ Inc,42,M,1 +7,1200.00,San Francisco,ABC Corp,55,F,0 +8,900.00,Los Angeles,XYZ Inc,37,M,0 +9,2500.00,Chicago,ABC Corp,33,F,1 +10,1800.00,New York,XYZ Inc,48,M,0 +11,750.00,San Francisco,ABC Corp,29,F,0 +12,2200.00,Chicago,XYZ Inc,51,M,0 +13,900.00,New York,ABC Corp,40,F,0 +14,1600.00,Los Angeles,XYZ Inc,26,M,0 +15,3000.00,San Francisco,ABC Corp,45,F,1 +16,1200.00,Chicago,XYZ Inc,34,M,0 +17,800.00,New York,ABC Corp,47,F,0 +18,1900.00,Los Angeles,XYZ Inc,32,M,0 +19,1100.00,San Francisco,ABC Corp,52,F,0 +20,4000.00,Chicago,XYZ Inc,38,M,1 +21,900.00,New York,ABC Corp,31,F,0 +22,1700.00,Los Angeles,XYZ Inc,49,M,0 +23,1000.00,San Francisco,ABC Corp,36,F,0 +24,2300.00,Chicago,XYZ Inc,27,M,1 +25,950.00,New York,ABC Corp,41,F,0 +26,1400.00,Los Angeles,XYZ Inc,54,M,0 +27,2800.00,San Francisco,ABC Corp,39,F,1 +28,1100.00,Chicago,XYZ Inc,44,M,0 +29,750.00,New York,ABC Corp,30,F,0 +30,2000.00,Los Angeles,XYZ Inc,46,M,0 +31,1250.00,San Francisco,ABC Corp,35,F,0 +32,2100.00,Chicago,XYZ Inc,43,M,0 +33,950.00,New York,ABC Corp,56,F,0 +34,1800.00,Los Angeles,XYZ Inc,29,M,0 +35,3200.00,San Francisco,ABC Corp,48,F,1 +36,1300.00,Chicago,XYZ Inc,37,M,0 +37,900.00,New York,ABC Corp,51,F,0 +38,2000.00,Los Angeles,XYZ Inc,33,M,0 +39,1050.00,San Francisco,ABC Corp,42,F,0 +40,2400.00,Chicago,XYZ Inc,26,M,0 +41,800.00,New York,ABC Corp,45,F,0 +42,1500.00,Los Angeles,XYZ Inc,31,M,0 +43,2800.00,San Francisco,ABC Corp,50,F,1 +44,1350.00,Chicago,XYZ Inc,28,M,0 +45,920.00,New York,ABC Corp,47,F,0 +46,2000.00,Los Angeles,XYZ Inc,36,M,0 +47,1125.00,San Francisco,ABC Corp,52,F,0 +48,1900.00,Chicago,XYZ Inc,38,M,1 +49,850.00,New York,ABC Corp,32,F,0 +50,1750.00,Los Angeles,XYZ Inc,49,M,0 +51,950.00,San Francisco,ABC Corp,27,F,0 +52,2300.00,Chicago,XYZ Inc,41,M,0 +53,850.00,New York,ABC Corp,54,F,0 +54,1600.00,Los Angeles,XYZ Inc,39,M,0 +55,3000.00,San Francisco,ABC Corp,46,F,1 +56,1250.00,Chicago,XYZ Inc,35,M,0 +57,800.00,New York,ABC Corp,56,F,0 +58,2200.00,Los Angeles,XYZ Inc,29,M,0 +59,1050.00,San Francisco,ABC Corp,48,F,0 +60,4000.00,Chicago,XYZ Inc,37,M,1 +61,950.00,New York,ABC Corp,30,F,0 +62,1700.00,Los Angeles,XYZ Inc,49,M,0 +63,1000.00,San Francisco,ABC Corp,36,F,0 +64,2800.00,Chicago,XYZ Inc,27,M,1 +65,900.00,New York,ABC Corp,41,F,0 +66,1400.00,Los Angeles,XYZ Inc,54,M,0 +67,3200.00,San Francisco,ABC Corp,39,F,1 +68,1100.00,Chicago,XYZ Inc,44,M,0 +69,750.00,New York,ABC Corp,30,F,0 +70,2000.00,Los Angeles,XYZ Inc,46,M,0 +71,1250.00,San Francisco,ABC Corp,35,F,0 +72,2100.00,Chicago,XYZ Inc,43,M,0 +73,950.00,New York,ABC Corp,56,F,0 +74,1800.00,Los Angeles,XYZ Inc,29,M,0 +75,3200.00,San Francisco,ABC Corp,48,F,1 +76,1300.00,Chicago,XYZ Inc,37,M,0 +77,900.00,New York,ABC Corp,51,F,0 +78,2000.00,Los Angeles,XYZ Inc,33,M,0 +79,1050.00,San Francisco,ABC Corp,42,F,0 +80,2400.00,Chicago,XYZ Inc,26,M,0 +81,800.00,New York,ABC Corp,45,F,0 +82,1500.00,Los Angeles,XYZ Inc,31,M,0 +83,2800.00,San Francisco,ABC Corp,50,F,1 +84,1350.00,Chicago,XYZ Inc,28,M,0 +85,920.00,New York,ABC Corp,47,F,0 +86,2000.00,Los Angeles,XYZ Inc,36,M,0 diff --git a/zavrazhnova_svetlana_lab_6/zavrazhnova_svetlana_lab_6.py b/zavrazhnova_svetlana_lab_6/zavrazhnova_svetlana_lab_6.py new file mode 100644 index 0000000..b59d937 --- /dev/null +++ b/zavrazhnova_svetlana_lab_6/zavrazhnova_svetlana_lab_6.py @@ -0,0 +1,84 @@ +# import pandas as pd +# from sklearn.neural_network import MLPClassifier +# from sklearn.preprocessing import OneHotEncoder +# +# # Загрузка данных из файла +# df = pd.read_csv('fraud_dataset.csv') +# +# # Разделение признаков и целевой переменной +# # В данном случае целевая переменная - fraud_label +# X = df[['transaction_amount', 'location', 'merchant', 'age', 'gender']] +# y = df['fraud_label'] +# +# X[['location', 'merchant', 'gender']] = X[['location', 'merchant', 'gender']].astype(str) +# +# # Преобразование категориальных признаков в числовые +# enc = OneHotEncoder() +# X_encoded = enc.fit_transform(X[['location', 'merchant', 'gender']]).toarray() +# +# # Объединение закодированных признаков с числовыми признаками +# X_final = pd.concat([X['transaction_amount'], pd.DataFrame(X_encoded), X['age']], axis=1) +# +# # Создание модели MLPClassifier +# model = MLPClassifier(hidden_layer_sizes=(100, 100)) # Можно настроить размеры и количество скрытых слоев +# +# # Обучение модели +# model.fit(X_final, y) +# +# # Предсказание для новых данных +# # Пример предсказания для новой транзакции +# new_transaction = pd.DataFrame([[2000.00, 'New York', 'ABC Corp', 35, 'M']], +# columns=['transaction_amount', 'location', 'merchant', 'age', 'gender']) +# new_transaction_encoded = enc.transform(new_transaction[['location', 'merchant', 'gender']]).toarray() +# new_transaction_final = pd.concat([new_transaction['transaction_amount'], +# pd.DataFrame(new_transaction_encoded), +# new_transaction['age']], axis=1) +# prediction = model.predict(new_transaction_final) +# +# # Вывод предсказания +# print(prediction) + +import pandas as pd +import numpy as np +from matplotlib import pyplot as plt +from sklearn import metrics +from sklearn.model_selection import train_test_split +from sklearn.neural_network import MLPRegressor +from sklearn.preprocessing import LabelEncoder + +# Метод обучения нейронной сети +def reg_neural_net(): + df = pd.read_csv('fraud_dataset.csv') + x, y = [df.drop("fraud_label", axis=1).values, + df["fraud_label"].values] + encoder = LabelEncoder() + df['location'] = encoder.fit_transform(df['location']) # Преобразование категориального столбца "location" в числовой формат + + x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.001, random_state=42) + + mlp = MLPRegressor(hidden_layer_sizes=(100, 50), activation='tanh', solver='adam', random_state=15000) + mlp.fit(x_train, y_train) + y_predict = mlp.predict(x_test) + err = pred_errors(y_predict, y_test) + make_plots(y_test, y_predict, err[0], err[1], "Нейронная сеть") + + +# Метод рассчёта ошибок +def pred_errors(y_predict, y_test): + mid_square = np.round(np.sqrt(metrics.mean_squared_error(y_test, y_predict)),3) # Рассчёт среднеквадратичной ошибки модели + det_kp = np.round(metrics.r2_score(y_test, y_predict), 2) # Рассчёт коэфициента детерминации модели + return mid_square, det_kp + + +# Метод отрисовки графиков +def make_plots(y_test, y_predict, mid_sqrt, det_kp, title): + plt.plot(y_test, c="red", label="\"y\" исходная") # Создание графика исходной функции + plt.plot(y_predict, c="green", label="\"y\" предсказанная \n" + "Ср^2 = " + str(mid_sqrt) + "\n" + "Кд = " + str(det_kp)) # Создание графика предсказанной функции + plt.legend(loc='lower left') + plt.title(title) + plt.savefig('static/' + title + '.png') + plt.close() + +reg_neural_net() \ No newline at end of file