def f_trainModelMain(train_path, test_path, index_name, target_name, userPath, modeltype, arithmetic):
'''
调用模型的主函数
:param train_path: 训练集 str DataFrame
:param test_path: 测试集 str DataFrame
:param index_name: 索引
:param target_name: 目标变量列名
:param userPath: 用户路径
:param modeltype: 模型的简称 有12个取值 [ccxboost,ccxgbm,ccxrf] + [demo,speed,accuracy,stable]
:param arithmetic: [Xgboost,GBM,RF]
:return: 最优模型结果的文件路径
'''
modelCode = f_genmodelCodeDict(userPath)
modelmain = ModelMain(train_path, test_path, index_name, target_name)
if arithmetic == 'Xgboost':
return f_xgboost(modelmain, modeltype, modelCode)
elif arithmetic == 'GBM':
return f_gbm(modelmain, modeltype, modelCode)
elif arithmetic == 'RF':
return f_rf(modelmain, modeltype, modelCode)
else:
# 写日志
print('错误码003 模型没有跑起来')
return []
def f_trainModelMain(train_path, test_path, index_name, target_name, userPath, modeltype, arithmetic,optimizationType,is_auto,param):
'''
调用模型的主函数
:param train_path: 训练集 str DataFrame
:param test_path: 测试集 str DataFrame
:param index_name: 索引
:param target_name: 目标变量列名
:param userPath: 用户路径
:param modeltype: 模型的简称 有24个取值 [grid,bayes] + [ccxboost,ccxgbm,ccxrf] + [demo,speed,accuracy,stable]
:param arithmetic: [Xgboost,GBM,RF]
:param optimizationType: [grid,bayes]
:param is_auto: 是否自动寻参
:param param: 自动寻参的超参数范围
:return: 最优模型结果的文件路径
'''
modelCode = f_genmodelCodeDict(userPath)
modelmain = ModelMain(train_path, test_path, index_name, target_name)
#自动寻优的入口 is_auto==0 表示手动 仅仅bayes方式存在手动调参
if int(is_auto) == 0 and optimizationType == 'bayes':
try:
#进来了 下面就是自动寻参的天下了
manual_conf_path = create_conf(arithmetic, param,userPath)
if arithmetic == 'Xgboost':
return f_recursionboostModel(train_path, test_path, index_name, target_name, manual_conf_path, 0,
optimizationType)
elif arithmetic == 'GBM':
return f_recursiongbmModel(train_path, test_path, index_name, target_name, manual_conf_path, 0,
optimizationType)
elif arithmetic == 'RF':
return f_recursionrfModel(train_path, test_path, index_name, target_name, manual_conf_path, 0,
optimizationType)
else:
# 写日志
print('错误码005 模型没有跑起来')
return []
except Exception as e:
print('错误码004 手动寻参出错问题为：')
print("ErrorMsg:\t"+str(e))
return []
if arithmetic == 'Xgboost':
return f_xgboost(modelmain, modeltype, modelCode,optimizationType)
elif arithmetic == 'GBM':
return f_gbm(modelmain, modeltype, modelCode,optimizationType)
elif arithmetic == 'RF':
return f_rf(modelmain, modeltype, modelCode,optimizationType)
else:
# 写日志
print('错误码003 模型没有跑起来')
return []
def f_recursionrfModel(train_path, test_path, index_name, target_name, modelconf, i,optimizationType):
'''
递归的将上一轮的重要变量重新作为输入 从而达到筛选变量的作用
:param train_path: 训练集
:param test_path: 测试集
:param index_name:
:param target_name:
:param modelconf: 模型配置文件路径
:param i: 记录递归的次数
:return: 最终递归完成的模型输出 结果的路径列表
'''
train_path = ModelUtil.load_data(train_path)
test_path = ModelUtil.load_data(test_path)
initmodelmain = ModelMain(train_path, test_path, index_name, target_name)
initpathlist = initmodelmain.ccxrf_main(modelconf,optimizationType)
# 1.计算出重要变量的个数
implen, impvar = f_getImplen(initpathlist[2])
# 2.计算出模型的AUC和KS
train_auc, train_ks = f_getAucKs(initpathlist[3])
test_auc, test_ks = f_getAucKs(initpathlist[4])
# 3.判断出模型重要变量占总变量的百分比情况
imppct = f_getVarpctrf(initpathlist[1], implen) # 入模变量 == 重要变量
flag = f_flag(train_auc, train_ks, test_auc, test_ks, imppct)
i = i + 1
if i < 5:
if flag:
print('递归调用 ' * 20)
newselectcol = impvar + [index_name, target_name]
print('---入选模型的变量个数%s' % len(newselectcol))
train_path = ModelUtil.load_data(train_path)[newselectcol]
test_path = ModelUtil.load_data(test_path)[newselectcol]
print('##' * 20, i, '##' * 20)
# 后续优化 递归的同时修改配置文件modelconf
return f_recursionrfModel(train_path, test_path, index_name, target_name, modelconf, i,optimizationType)
else:
print('满足条件结束递归 ' * 10)
return initpathlist
else:
print('递归次数达到要求结束递归' * 10)
return initpathlist
r'C:\Users\liyin\Desktop\CcxFpABSModel\updateModel\FPALLVAR_1103.csv')
train = pd.merge(train_index, FPALLVAR)
test = pd.merge(test_index, FPALLVAR)
######
train_path = r'C:\Users\liyin\Desktop\CcxFpABSModel\updateModel\train_1103.csv' # 训练集数据路径，可见demo_data文件夹
test_path = r'C:\Users\liyin\Desktop\CcxFpABSModel\updateModel\test_1103.csv' # 测试集数据路径，可见demo_data文件夹
# train.to_csv(train_path, index=False)
# test.to_csv(test_path, index=False)
############
from ccxmodel.modelmain import ModelMain
index_name = 'lend_request_id' # 数据集唯一索引，有且仅支持一个索引，不支持多个索引
target_name = 'TargetBad_P12' # 目标变量
modelmain = ModelMain(train_path, test_path, index_name, target_name)
modelmain.ccxboost_main(
r'C:\Users\liyin\Desktop\CcxFpABSModel\updateModel\ccxboost_1.conf')
r"""
{'colsample_bytree': 0.80000000000000004, 'eta': 0.10000000000000001, 'eval_metric': 'auc', 'gamma': 2, 'lambda': 500, 'max_depth': 4, 'max_test_auc': 0.71066600000000002, 'max_train_auc': 0.75396439999999987, 'min_child_weight': 2, 'num_best_round': 144, 'num_boost_round': 500, 'num_maxtest_round': 494, 'objective': 'binary:logistic', 'subsample': 0.80000000000000004, 'gap': 0.042999999999999997}
模型保存成功 文件路径名：C:\Users\liyin\Desktop\CcxFpABSModel\updateModel\model20171103214623/modeltxt/model_Fp_Ccx_All_2017-11-03.txt
重要变量的个数：145
数据保存成功:C:\Users\liyin\Desktop\CcxFpABSModel\updateModel\model20171103214623/modeldata/d_2017-11-03_importance_var.csv
训练集模型报告：
precision recall f1-score support
0.0 0.86 1.00 0.92 12354
1.0 0.77 0.02 0.04 2068
avg / total 0.85 0.86 0.80 14422
测试集模型报告：