逻辑回归简单应用

import csv
import pandas as pd
from sklearn.linear_model import LogisticRegression

import warnings
warnings.filterwarnings('ignore') #忽视

# 加载数据源,并etl
loans_2020 = pd.read_csv('/Users/yee/Desktop/LoanStats3a.csv', skiprows=1) #第一行是字符串，所以要skiprows=1跳过第一行
half_count = len(loans_2020) / 2 # 4万行除以2 = 19767.5行
loans_2020 = loans_2020.dropna(thresh=half_count, axis=1)#2万行中剔除空白值超过一半的列，thresh：剔除
loans_2020 = loans_2020.drop(['desc', 'url'],axis=1) #按照列中，删除描述和URL链接
print("原始数据 = ",loans_2020.shape)

#删除不需要的列，只需要特征和标签
loans_2020 = loans_2020.drop(
["id", "member_id", "funded_amnt", "funded_amnt_inv", "grade", "sub_grade", "emp_title", "issue_d",
"zip_code", "out_prncp", "out_prncp_inv", "total_pymnt", "total_pymnt_inv", "total_rec_prncp",
"total_rec_int", "total_rec_late_fee", "recoveries", "collection_recovery_fee", "last_pymnt_d", "last_pymnt_amnt"]
, axis=1)
print(loans_2020.iloc[1])#第一行数据
print("现在的列数 = ",loans_2020.shape[1 ])

#标签列
#Fully Paid：批准了客户的贷款，后面给他打个“1”
#Charged Off：没有批准了客户的贷款，后面给他打个“0”
#Does not meet the credit policy. Status:Fully Paid：，没有满足要求的有1988个，也不要说清楚不贷款，就不要这个属性了
#Late (16-30 days)  ：延期了16-30 days
#Late (31-120 days)：延期了31-120 days ， 所以这些都不确定的属性，相当于“取保候审”
print(loans_2020['loan_status'].value_counts())#计算该列特征的属性的个数
loans_2020 = loans_2020[(loans_2020['loan_status'] == "Fully Paid") |
                        (loans_2020['loan_status'] == "Charged Off")]
status_replace = {
    "loan_status": {
        "Fully Paid": 1, #完全支付
        "Charged Off": 0 #违约
    }
}
loans_2020 = loans_2020.replace(status_replace)

#值都是一样的列没有意义，删除这些列
orig_columns = loans_2020.columns  #展现出所有的列
drop_columns = []  #需要删除的列放在此列表
for col in orig_columns:
    col_series = loans_2020[col].dropna().unique()  # 对每一列的数据去空去重
    if len(col_series) == 1:  # 如果该特征的属性只有一个属性，就给过滤掉该特征
        drop_columns.append(col)
loans_2020 = loans_2020.drop(drop_columns, axis=1)
print(loans_2020.shape)

#对存在None的列进行分析，不要这一列或者删除None的那一行
loans = loans_2020
null_counts = loans.isnull().sum()  #统计一下每列的缺失值有多少
print("每列的缺失值",null_counts)
loans = loans.drop("pub_rec_bankruptcies", axis=1) #因为None的太多，这列直接删除
loans = loans.dropna(axis=0) #其他为None的删除行
print(loans_2020.shape)

#由于sklearn库不接受字符型的数据，所以还需将上面特征中12个字符型的数据进行处理。
object_columns_df = loans.select_dtypes(include=["object"])
print(object_columns_df.iloc[0])

#删除：last_credit_pull_d：LC撤回最近的月份
#删除：earliest_cr_line：第一次借贷时间
#删除：addr_state：家庭邮编
#删除：title：URL的标题
loans = loans.drop(
    ["last_credit_pull_d", "earliest_cr_line", "addr_state", "title"], axis=1)

#int_rate：利息，10.65%，后面还要把%去掉
loans["int_rate"] = loans["int_rate"].str.rstrip("%").astype("float")
#revol_util：透支额度占信用比例
loans["revol_util"] = loans["revol_util"].str.rstrip("%").astype("float")
#emp_length：10年的映射成10，9年的映射成9
mapping_dict = {
    "emp_length": {
        "10+ years": 10,
        "9 years": 9,
        "8 years": 8,
        "7 years": 7,
        "6 years": 6,
        "5 years": 5,
        "4 years": 4,
        "3 years": 3,
        "2 years": 2,
        "1 year": 1,
        "< 1 year": 0,
        "n/a": 0
    }
}
loans = loans.replace(mapping_dict)

print("-----------------------------特征工程：对这些列的数据行转列，打上0/1的独热标签----------------------------------------")
#home_ownership：房屋所有权
#verification_status：身份保持证明
#emp_length：客户公司名称
#purpose：贷款的意图
#term：贷款分期的时间
cat_columns = ["home_ownership", "verification_status", "emp_length", "purpose", "term"]
# 哑变量
dummy_df = pd.get_dummies(loans[cat_columns])
loans = pd.concat([loans, dummy_df], axis=1) #连接两个或多个数组,
loans = loans.drop(cat_columns, axis=1)
#pymnt_plan 指示是否已为贷款实施付款计划 ，里面都为N，删掉这一列
loans = loans.drop("pymnt_plan", axis=1)
print(loans.iloc[0])

print("-----------------------------使用逻辑回归算法库---------------------------------------")
cols = loans.columns #所有列
train_cols = cols.drop("loan_status") # 删除loan_status这一列作为目标值
features = loans[train_cols]  # 特征矩阵
target = loans["loan_status"] # 作为标签矩阵
lr = LogisticRegression() # 调用逻辑回归的算法包
lr.fit(features, target) #开始训练
predictions = lr.predict(features) # 开始预测,结果是个ndarray
proba = lr.predict_proba(features)
print(proba[:10]) #前十行模型算出的值，<=0.5为0，>0.5为1
print(predictions[:10])#0:代表不会偿还  1:代表偿还

print("-----------------------------验证模型的好坏---------------------------------------")
#用4个指标 fp tp fn tn，统计四种情况的数量
# 假正类（False Positive，FP）：将负类预测为正类
fp_filter = (predictions == 1) & (loans["loan_status"] == 0)
fp = len(predictions[fp_filter])
print(fp)
print("----------------------------------------")

# 真正类（True Positive，TP）：将正类预测为正类
tp_filter = (predictions == 1) & (loans["loan_status"] == 1)
tp = len(predictions[tp_filter])
print(tp)
print("----------------------------------------")

# 假负类（False Negative，FN）：将正类预测为负类
fn_filter = (predictions == 0) & (loans["loan_status"] == 1)
fn = len(predictions[fn_filter])
print(fn)
print("----------------------------------------")

# 真负类（True Negative，TN）：将负类预测为负类
tn_filter = (predictions == 0) & (loans["loan_status"] == 0)
tn = len(predictions[tn_filter])
print(tn)

#真正率：TPRate的意义是所有真实类别为1的样本中，预测类别为1的比例。越高越好！
#假正率：FPRate的意义是所有真实类别为0的样本中，预测类别为1的比例。越低越好！
tpr = tp / float((tp + fn))
fpr = fp / float((fp + tn))
print(tpr)#真正率
print(fpr)#假正率

print("-------------------改变权重重新计算--------------------------------------------------")
"""
权重项可以自己定义的
0代表5倍的
"""
from sklearn.model_selection import cross_val_predict
penalty = {
    0: 5,
    1: 1
}

lr = LogisticRegression(class_weight=penalty)

predictions = cross_val_predict(lr, features, target, cv=10)
predictions = pd.Series(predictions)

# 建立混淆矩阵
# False positives.
fp_filter = (predictions == 1) & (loans["loan_status"] == 0)
fp = len(predictions[fp_filter])

# True positives.
tp_filter = (predictions == 1) & (loans["loan_status"] == 1)
tp = len(predictions[tp_filter])

# False negatives.
fn_filter = (predictions == 0) & (loans["loan_status"] == 1)
fn = len(predictions[fn_filter])

# True negatives
tn_filter = (predictions == 0) & (loans["loan_status"] == 0)
tn = len(predictions[tn_filter])

# Rates
tpr = tp / float((tp + fn))
fpr = fp / float((fp + tn))

print(tpr)
print(fpr)