python交易预警_基于Python的客户流失预警算法

import pandas as pd

import seaborn as sns

import matplotlib.pyplot as plt

from sklearn.preprocessing import StandardScaler

train_df = pd.read_csv('训练集所在路径')

test_df = pd.read_csv('测试集所在路径')

full_df = pd.concat([train_df,test_df], ignore_index=True, sort=False)

print(full_df.tail())

#数据预处理

print(sns.heatmap(full_df.isnull(), cbar=False).set_title('Heatmap'))

full_df['CUST_AGE'].fillna(full_df['CUST_AGE'].mode()[0], inplace=True)

full_df['LOAN_PERIED'].fillna(method='backfill', inplace=True, limit=None)

full_df['LOAN_AMOUNT'].fillna(method='backfill', inplace=True, limit=None)

full_df['F_INSORNOT'].fillna(method='backfill', inplace=True, limit=None)

full_df['ALL_BUYINS_N'].fillna(method='backfill', inplace=True, limit=None)

full_df['DLRSI_CNT'].fillna(method='backfill', inplace=True, limit=None)

full_df['GLASSBUYSEPARATE_CNT'].fillna(method='backfill', inplace=True, limit=None)

full_df['SII_CNT'].fillna(method='backfill', inplace=True, limit=None)

full_df['CAR_COLOR'].fillna('NA', inplace=True, limit=None)

full_df['CUST_MARRY'].fillna(0, inplace=True)

full_df['CUST_MARRY'].replace('未婚', 0, inplace=True)

full_df['CUST_MARRY'].replace('已婚', 1, inplace=True)

print(full_df.isnull().sum())

#保存预处理完的训练集和测试集

train_clean = full_df[full_df['IS_LOST'].notnull()]

train_clean.to_csv('train_clean.csv')

test_clean = full_df[full_df['IS_LOST'].isnull()]

test_clean.to_csv('test_clean.csv')

# 开始建模

cardata = pd.read_csv('train_clean的路径')

cardatavar = cardata

scaler = StandardScaler(copy=False)

scaler.fit_transform(cardatavar[['CUST_AGE', 'CAR_AGE', 'CAR_PRICE', 'LOAN_AMOUNT']])

cardatavar[['CUST_AGE', 'CAR_AGE', 'CAR_PRICE', 'LOAN_AMOUNT']] = scaler.transform(cardatavar[['CUST_AGE', 'CAR_AGE', 'CAR_PRICE', 'LOAN_AMOUNT']])

plt.figure(figsize=(8, 4))

numbox = sns.boxplot(data=cardatavar[['CUST_AGE', 'CAR_AGE', 'CAR_PRICE', 'LOAN_AMOUNT']], palette="Set2")

plt.title("Check outliers")

plt.show()

cardatavar.to_csv('cardatavar.csv')

#标签分类

feature_cols = ['CUST_SEX', 'CUST_AGE', 'CUST_MARRY', 'BUYERPART', 'CAR_AGE',

'CAR_PRICE', 'IS_LOAN','LOAN_PERIED', 'LOAN_AMOUNT', 'F_INSORNOT', 'ALL_BUYINS_N', 'DLRSI_CNT',

'GLASSBUYSEPARATE_CNT', 'SII_CNT']

X = cardatavar[feature_cols]

y = cardatavar.IS_LOST

#逻辑斯蒂回归算法

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.05, random_state=0)

from sklearn.linear_model import LogisticRegression

logreg = LogisticRegression(solver='newton-cg')

logreg.fit (X_train, y_train)

y_pred = logreg.predict(X_test)

from sklearn import metrics

cnf_matrix = metrics.confusion_matrix(y_test, y_pred)

print(cnf_matrix)

print('Accuracy:{:.2f}'.format(metrics.accuracy_score(y_test, y_pred)))

print('Precision:{:.2f}'.format(metrics.precision_score(y_test, y_pred)))

print('Recall:{:.2f}'.format(metrics.recall_score(y_test, y_pred)))

print('F1:{:.2f}'.format(metrics.f1_score(y_test, y_pred)))

#神经网络算法

from sklearn.datasets import load_wine

wine = load_wine()

X, y = wine.data, wine.target

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.5, random_state=0)

from sklearn. preprocessing import StandardScaler

scaler = StandardScaler() . fit(X_train)

X_train = scaler. transform(X_train)

X_test = scaler. transform(X_test)

from sklearn.neural_network import MLPClassifier

model = MLPClassifier(solver="lbfgs", hidden_layer_sizes=(100,))

model. fit(X_train, y_train)

y_predict_on_train = model .predict(X_train)

y_predict_on_test = model. predict(X_test)

from sklearn. metrics import accuracy_score

print('训练集的准确率为: {: .2f}%'.format(100 * accuracy_score(y_train, y_predict_on_train)))

print('测试集的准确率为: {: .2f}%'.format(100 * accuracy_score(y_test, y_predict_on_test)))