lzh
184 lines
7.0 KiB
Python
184 lines
7.0 KiB
Python
import json
|
||
import os
|
||
|
||
from bson import json_util
|
||
|
||
"""
|
||
流程：
|
||
` 1、从monodb读取用户上传的数据集（已完成）`
|
||
`2、传入需要处理的列(特征列)，以及处理过程函数。
|
||
* 内置填充缺失值、删除缺失值以及one-hot编码
|
||
- 标准化数据（MinMax），可选参数 `
|
||
` 3、模型构建，传入模型名称、目标列进行构建`
|
||
- 允许用户进行参数搜索，前端用户输入的搜索参数格式为json格式
|
||
|
||
主函数生成：
|
||
- 自动生成代码的代码模块存放于同一个文件夹：
|
||
仅包含功能函数
|
||
|
||
- 主函数预先定义一个代码文件，相关参数通过占位符填充，填充的参数来源于前段输入，包括：主要包括特征列、目标列，文件名
|
||
"""
|
||
from utils.MODEL_DICT import MODEL_DICT
|
||
from ModelSelection.dataset_process_model import joint_code
|
||
from utils.mongodb_util import MongoUtil
|
||
|
||
|
||
class SetModel:
|
||
"""
|
||
|
||
用于与前端界面交互，获取特征列，以及数据处理步骤。
|
||
根据用户选择的步骤，读取预定义的代码。
|
||
前端返回参数
|
||
{
|
||
target:[],
|
||
features:[],
|
||
|
||
}
|
||
"""
|
||
|
||
def __init__(self, **params):
|
||
"""
|
||
|
||
:param name(str,):任务名称
|
||
:param dataset_name(str,):数据集名称
|
||
:param features(str of list):特征列
|
||
:param target(str):目标
|
||
:param model_type(str):分类/回归/聚类
|
||
:param model_name(str of list):模型
|
||
:param evaluate_methods(str of list,非必填):模型评估方法
|
||
:param username(str)
|
||
:param model_scene(str):模型使用场景
|
||
"""
|
||
self.name = params.get("name", "")
|
||
self.dataset_name = params.get("dataset_name", "")
|
||
self.target = params.get("target", "")
|
||
self.features = params.get("features", [])
|
||
self.model_type = params.get("model_type", "")
|
||
self.model_name = params.get("model_name", [])
|
||
self.evaluate_methods = params.get("metrics", [])
|
||
self.username = params.get("username", "")
|
||
self.model_scene = params.get("model_scene", "")
|
||
self.generate = ''
|
||
self.clean_code = ''
|
||
|
||
def get_code(self):
|
||
"""
|
||
根据模板，生成代码
|
||
:return:
|
||
"""
|
||
try:
|
||
# 拼接模型需要的库
|
||
for model in self.model_name:
|
||
self.generate += MODEL_DICT[self.model_type][model] + "\n"
|
||
# 拼接导入的库,并填充读取的文件名
|
||
self.generate += joint_code('ImportPackages.py') % self.dataset_name.replace('_', '.')
|
||
# 拼接变量
|
||
self.generate += "\nFEATURES={}\nTARGET='{}'".format(self.features, self.target)
|
||
# 拼接调用的模型
|
||
sklearn_models = [MODEL_DICT[self.model_type].get(model, " ").split(' ')[-1] + '()' for model in self.model_name]
|
||
self.generate += "\nMODEL = [{}]".format(", ".join(sklearn_models))
|
||
|
||
# 拼接分类/回归/聚类的主函数与必要评估方法
|
||
if self.model_type == "分类":
|
||
self.generate += joint_code('main_supervisied.py')
|
||
self.generate += joint_code("evaluation_classifier.py")
|
||
elif self.model_type == "回归":
|
||
self.generate += joint_code('main_supervisied.py')
|
||
self.generate += joint_code("evaluation_regressor.py")
|
||
elif self.model_type == "聚类":
|
||
self.generate += joint_code("main_unsupervised.py")
|
||
self.generate += joint_code("evaluation_cluster.py")
|
||
|
||
# 拼接用户自选的可视化的评估方法
|
||
if len(self.evaluate_methods) != 0:
|
||
for method in self.evaluate_methods:
|
||
location = MODEL_DICT.get(method)
|
||
if location:
|
||
self.generate += joint_code(location)
|
||
|
||
# 生成代码文件
|
||
filename = "generate_{}_{}.py".format(self.username, self.name)
|
||
filepath = os.path.join(os.path.abspath(''), 'temp', filename)
|
||
print("存放路径:", filepath)
|
||
with open(filepath, 'w', encoding='utf-8') as f:
|
||
f.write(self.generate)
|
||
except Exception as e:
|
||
self.generate = str(e)
|
||
# 返回生成代码的文本
|
||
return self.generate
|
||
|
||
def save_params(self, hide_history=True):
|
||
"""
|
||
保存当前参数(如果不重复的话)到数据库。
|
||
hide_history:bool:是否向用户展示该条历史记录，
|
||
hide_history默认为True，当用户运行成功当前代码时自动保存，
|
||
该条历史记录保存在admin用户中，该记录用于后续推荐。
|
||
当用户在界面上选择保存当前参数时，hide_history为False，正常保存
|
||
:return:
|
||
"""
|
||
try:
|
||
params = self.__dict__
|
||
db_util = MongoUtil()
|
||
del params["generate"], params["clean_code"]
|
||
check_duplicate_query = {
|
||
"username": self.username if not hide_history else "admin",
|
||
"name": self.name
|
||
}
|
||
exists = db_util.find_object("history_model", check_duplicate_query)
|
||
# 不存在这条记录且正常登陆，将这条记录加入历史记录集合
|
||
if not exists and self.username:
|
||
db_util.insert_object("history_model", params)
|
||
# 存在则更新记录
|
||
elif exists and self.username:
|
||
db_util.update_object("history_model", check_duplicate_query, params)
|
||
return True
|
||
except Exception as e:
|
||
raise e
|
||
|
||
def get_history(self):
|
||
"""
|
||
获取用户的所有历史记录,将所有列表项转换成,分割
|
||
:return:
|
||
"""
|
||
try:
|
||
db_util = MongoUtil()
|
||
search_query = {
|
||
"username": self.username,
|
||
}
|
||
history_list = db_util.find_object("history_model", search_query)
|
||
history_list = json.loads(json_util.dumps(history_list))
|
||
for index, history in enumerate(history_list):
|
||
for k, v in history.items():
|
||
if v is None:
|
||
history_list[index][k] = ""
|
||
if isinstance(v, list):
|
||
history_list[index][k] = ",".join(v)
|
||
return history_list
|
||
except Exception as e:
|
||
raise e
|
||
|
||
def delte_history(self):
|
||
"""
|
||
删除历史记录
|
||
:param object:dict 删除对象的查询条件
|
||
:return:
|
||
"""
|
||
try:
|
||
db_util = MongoUtil()
|
||
db_util.delete_object("history_model", {
|
||
"username": self.username,
|
||
"name": self.name
|
||
})
|
||
return True
|
||
except Exception as e:
|
||
raise e
|
||
|
||
|
||
if __name__ == '__main__':
|
||
myModel = SetModel(username="lzh3")
|
||
ll = myModel.get_history()
|
||
print(ll)
|
||
print("======================")
|
||
for i in ll:
|
||
print(i)
|