首页 | 本学科首页   官方微博 | 高级检索  
相似文献
 共查询到17条相似文献,搜索用时 46 毫秒
1.
本文目的是介绍第三种提高回归模型拟合优度的策略,即校正均值变换与其他变量变换。具体方法包括以下几个方面:①对多值名义自变量采取"校正均值变换";②对定量自变量引入派生变量,包括"对数变换""平方根变换""指数变换""平方变换""立方变换"和"交叉乘积变换"的结果;③对定量因变量分别采取"对数变换""平方根变换""指数变换""倒数变换"和"Logistic变换";④构建回归模型时,在假定"包含截距项"与"不含截距项"的条件下,分别采取"前进法""后退法"和"逐步法"筛选自变量。得到了如下结论:①对定量因变量和自变量不做变量变换时,回归模型的拟合优度非常低;②根据资料所具备的条件,对定量因变量采取不同的变量变换方法,其回归模型的拟合优度是不同的;③对多值名义自变量进行"校正均值变换"是合理的,且有助于提高回归模型拟合优度;④对定量自变量引入派生变量是非常有价值的;⑤假定回归模型中不含截距项有助于提高回归模型的拟合优度。  相似文献   

2.
本文目的是介绍第一种提高回归模型拟合优度的策略,即哑变量变换与其他变量变换。具体方法包括以下几个方面:①对多值名义自变量采取"哑变量变换";②对定量和有序自变量引入派生变量,包括"对数变换""平方根变换""指数变换""平方变换""立方变换"和"交叉乘积变换"的结果;③对定量因变量分别采取"对数变换""平方根变换""指数变换""倒数变换"和"Logistic变换";④构建回归模型时,在假定"包含截距项"与"不含截距项"的条件下,分别采取"前进法""后退法"和"逐步法"筛选自变量。得到了如下几个结论:①对定量因变量和自变量不做变量变换时,回归模型的拟合优度非常差;②根据资料所具备的条件,对定量因变量采取不同的变量变换方法,其回归模型的拟合优度是不尽相同的;③对多值名义自变量进行"哑变量变换"是常规的做法,但存在不足之处;④对定量自变量引入派生变量是非常有价值的;⑤假定回归模型中不含截距项有助于提高回归模型的拟合优度。  相似文献   

3.
本文目的是介绍第四种提高回归模型拟合优度的策略,即优化计分变换与其他变量变换。具体方法包括以下几个方面:①第一,对多值名义自变量采取"优化计分变换";②对有序自变量分别采取"单调变换"与"优化计分变换";③对定量自变量分别采取"样条变换"和"单调样条变换";④对定量因变量分别采取"样条变换""单调样条变换"和"BOX-COX变换"。全部变量变换方法组合起来共12种,共创建了12个多重非线性回归模型。依据"拟合优度评价指标"的取值,从12个回归模型中挑选出一个,即本文中的"模型1",其"均方误差平方根=0.30935、R~2=0.9586、调整R~2=0.9527"。结合本期科研方法专题同类文章的结果和结论,得出提高回归模型拟合优度的策略主要在于以下四点:①应对"定量因变量""定量自变量"和"多值有序自变量"采取合适的变量变换方法;②在拟合回归模型的过程中,应尽可能多地引入派生变量;③应假定回归模型中不含截距项;④在构建回归模型的过程中,应尽可能多地使用筛选自变量的策略,如"前进法""后退法"和"逐步法"。  相似文献   

4.
非配对设计二值资料一水平多重Logistic回归分析   总被引:1,自引:0,他引:1       下载免费PDF全文
本文的目的是介绍非配对设计二值资料一水平多重Logistic回归模型的构建与求解方法。基于SAS软件分别对以列联表和数据库形式呈现的定性资料进行全面分析,并得出了4个对提高模型拟合优度很有价值的结论:第一,若资料以列联表形式呈现,应拟合"加权"Logistic回归模型;第二,若资料中包含定量自变量,不适合将其定性化;第三,若资料中包含定量自变量,应依据定量自变量和二值自变量产生出派生自变量;第四,若资料中有定性自变量时,必须将多值名义或有序自变量进行哑变量变换,不需要依据二值自变量产生出派生自变量。  相似文献   

5.
本文目的是分析一个已知真实情况的资料,比较适应性回归分析与非适应性回归分析建模的效果。结论如下:当资料中存在与因变量确有关系的自变量时,ADAPTIVEREG过程具有较好的甄别能力;REG过程具有较好的甄别能力,但需要满足一定条件,即采用"前进法"或"逐步法"筛选自变量,同时还需要"假定模型包含截距项"。当资料中不存在与因变量确有关系的自变量时,ADAPTIVEREG过程几乎完全失去了甄别能力;REG过程具有较好的甄别能力,但需要满足一定条件,即采用"前进法"筛选自变量,同时还需要"假定模型包含截距项"。若研究者基于"基本常识"和"专业知识"确定的自变量都与因变量有关系,对因变量进行Logistic变换,并且,假定回归模型中不含截距项时,会在回归模型中保留非常多的自变量。  相似文献   

6.
本文目的是介绍回归建模的基础与要领之三,即"变量状态与相互间关系"。首先,介绍"因变量状态"与"自变量状态";其次,介绍"自变量间相互关系",即"自变量间相互独立""自变量间有线性关系"和"自变量间有非线性关系";最后,介绍"自变量与因变量间关系",包括"自变量与因变量间无任何数量关系""自变量与因变量间有间接数量关系"和"自变量与因变量间有直接数量关系"。很明显,清楚"变量状态和变量间关系"是构建合理回归模型的重要基础与要领之一。  相似文献   

7.
本文的目的是介绍m∶n配对设计二值资料一水平多重Logistic回归分析方法。首先,介绍了需要了解的基本概念;其次,介绍了构建此类回归模型的基本原理;最后,通过一个实例介绍了使用SAS实现计算的全过程。在此过程中,获得了如下四点启示:其一,有必要确保所获得的科研资料是值得分析的;其二,有必要基于定量自变量产生派生变量;其三,有必要同时采用"逐步法""前进法"和"后退法"筛选自变量;其四,有必要采用多种方法评价不同回归模型对资料的拟合优度。  相似文献   

8.
本文目的是介绍零膨胀Poisson分布模型回归分析。首先,介绍零膨胀计数资料及其零膨胀Poisson分布回归模型构建原理,包括"零膨胀Poisson分布回归模型的形式"和"零膨胀Poisson分布回归模型的求解";其次,介绍"零膨胀Poisson分布回归模型的SAS实现",包括"创建SAS数据集""呈现因变量Y的频数分布""求出因变量Y的均值和方差"和"基于全部自变量对因变量Y构建多重零膨胀Poisson分布回归模型"。本文结果提示,当计数资料为非严重过离散的零膨胀计数资料时,拟合"多重零膨胀Poisson分布回归模型",可获得满意的拟合效果。  相似文献   

9.
罗艳虹  胡良平 《四川精神卫生》2019,32(2):101-104,109
本文目的是通过分析一个带有8个噪声变量的数据集,揭示适应性回归模型的实际应用价值。在数据集包含两个自变量与因变量有密切数量联系的前提条件下,适应性回归模型受噪声变量的影响接近于零;在数据集包含一个自变量与因变量有密切数量联系的前提条件下,适应性回归模型受噪声变量的影响较大,其分析结果出现了一定程度的"失真";在数据集包含零个自变量与因变量有密切数量联系的前提条件下,适应性回归模型受噪声变量的影响非常大,其分析结果是完全不可信的。得出的结论是:适应性回归分析模型不是万能的,其结果的可信度取决于数据集中是否真正包含"客观存在的规律性"。  相似文献   

10.
本文目的是介绍主成分回归分析的概念、作用以及用软件实现计算的方法。先对自变量进行主成分分析,然后将主成分变量视为新的自变量,再进行多重线性回归分析。通过不引入和引入派生变量以及采取不同的策略筛选自变量,可以获得多个合格的多重线性回归模型。在回归模型自由度接近相等时,基于残差方差最小、复相关系数最大为评价指标,从众多回归模型中优中选优。得出的经验为:应慎用主成分回归分析。  相似文献   

11.
In order to obtain repeated measurements of depression in an efficient and relatively inexpensive design, a mixture of face-to-face interviews and mail questionnaires was employed. The aims of the study were to examine mode effects of face-to-face interviews versus mail questionnaires on depression scores and to test potential interactions between mode of data collection and sex and age of the respondents. In the study sample, which at the outset consisted of 327 depressed and 325 non-depressed older adults (55–85 years) drawn from a larger random community based sample in the Netherlands, depression was measured in successive waves (cycles), using the Center for Epidemiologic Studies of Depression scale (CES-D). With mode of data collection and sex and age of the respondents as independent variables, differences in CES-D scores were analysed. The CES-D scores were higher when collected by mail questionnaires than when face-to-face interviews were used. No systematic interactions between sex and age of the respondents with mode of data collection were found. For the scores based on mail questionnaires, a transformation is proposed, resulting in scores that are comparable to those obtained by interviews. In studying depression in older adults, more cost-effective mail questionnaires may be used in addition to face-to-face interviews, provided that a transformation is performed before embarking on the analysis. Copyright © 1999 Whurr Publishers Ltd.  相似文献   

12.

Objective

Low grade gliomas (LGGs) are slow-growing primary brain tumors with heterogeneous clinical behaviors. The aim of our study is to review the treatment outcome of 63 patients with LGGs focusing on surgical outcome and the current therapeutic strategy.

Methods

We retrospectively enrolled 63 patients surgically treated for LGGs. The gross total resection (GTR) was performed in 35 patients (60.3%), subtotal resection (STR) was performed in 19 patients (31.7%) and partial resection (PR) or biopsy was performed in 9 patients (14.3%). We analyzed their progression-free survival (PFS), overall survival (OS), and malignant transformation with regard to age, gender, Karnofsky performance score (KPS), clinical presentation, tumor location, radiologic pattern, contrast enhancement, extent of removal, pathologic subtype, chemotherapy (CT) and radiotherapy (RT) treatment.

Results

Among all LGGs, the 3-year OS rate was 80% and the 5-year OS was 76%. The 3-year PFS rate was 83.6% and the 5-year PFS was 25%. The non-eloquent area location showed a longer PFS than the eloquent area location (p = 0.05). Oligodendroglial pathology showed a longer PFS compared to oligoastrocytomas and astrocytomas (p = 0.02). Patients older than 60 years had poorer OS than younger patients (p < 0.05). Female gender had a shorter OS than male gender (p < 0.05), and a KPS of 90 or 100 had a longer OS than a KPS of 80 (p < 0.05). Oligodendroglial pathology statistically correlated with a longer OS (p < 0.05).

Conclusion

The findings from our study, which were confirmed by uni- and multivariate analyses, demonstrated that radical tumor resection was associated with better long-term outcomes and tumor progression for patients with LGG.  相似文献   

13.
本文目的是介绍方差齐性检验与SAS实现.方差齐性检验可分为以下三类:①直接基于方差比较的方差分析法;②对原始数据经过变量变换后的新数据采取基于均值比较的方差分析法;③对符合正态分布的定量原始数据采取χ2检验法.在第一类中,直接基于两样本方差比构造出服从F分布的检验统计量;在第二类中,对原始数据有多种不同的变量变换方法,对变换后的新数据进行单因素多水平设计一元定量资料方差分析;在第三类中,对服从正态分布的定量资料构造χ2检验统计量.本文基于SAS软件对三个实例进行方差齐性检验,并对输出结果作出解释.  相似文献   

14.
目的 探讨基于多模CT的区域软脑膜侧支评估(rLMC)与急性缺血性卒中梗死体积及出血转化的相关性及其应用价值.方法 回顾性分析2019年10月至2020年10月该院首次发病≤6h的急性大血管闭塞性缺血性脑卒中患者,颅脑CT排除出血性病变,采用rLMC分级,分为2个区域:大脑前动脉(ACA)-大脑中动脉(MCA)区和大脑后动脉(PCA)-MCA区.软脑膜动脉分级应用6分量表,CTA侧支评分为两个区域的总分(0~10分).入院3d内在头颅MR常规序列基础上加做DWI序列,明确梗死体积、是否合并出血转化.入院后10~14 d或患者病情加重时复查常规CT,了解是否有出血性转化.结果 不同rLMC分级评分组卒中家族史、冠心病史、入院时NIHSS评分、空腹血糖比较,差异有统计学意义(P<0.05).rLMC分级评分与脑梗死体积呈显著负相关(γ=-0.735,P<0.001).不同rLMC分级评分组在梗死体积、症状性颅内出血(sICH)发生率和颅内高密度灶(PCHDs)发生率比较,差异有统计学意义(P<0.05).多因素Logistic回归分析提示卒中家族史、血糖、入院时NIHSS评分是软脑膜侧支循环的影响因素.结论 rLMC与急性缺血性卒中梗死体积、出血转化呈显著负相关,可有效反映急性缺血性卒中的病情进展及严重程度.  相似文献   

15.
16.
Summary The rationale behind the evaluation of natural differentiating agents, such as nerve growth factor (NGF), for reverse transforming potential is based on the theory that such compounds may represent a nontoxic means of controlling tumor growth. Previous in vitro experiments have shown that NGF is capable of retarding growth and of inducing persistent differentiation of neurogenic tumor cell lines. In vivo, NGF is capable of causing a persistent reduction in the number of ethylnitrosourea-induced neurinomas and of increasing survival time following intracerebral implantation of F98 anaplastic glioma cells. In this study, anaplastic glioma and neurinoma implants were treated with NGF to evaluate the reverse transforming potential of NGF in vivo. Results indicate that NGF is capable of causing a significant decrease in the growth rate of subcutaneous T9 (anaplastic glioma) and clone 16 (anaplastic neurinoma) implants. Significantly, NGF treatment was accompanied by adverse effects that were minimal and transient. Continued tumor growth (although greatly retarded) following NGF treatment is an aspect that requires further investigation. However, the results of this study suggest that NGF may prove useful, alone or in combination with other types of therapy, for the treatment of tumors of neurogenic origin.Supported by NIH grant CA32594, an NIH Post Doctoral Fellowship to M. J. Y., a grant from the Preuss Foundation and a BRSG grant (College of Veterinary Medicine, Michigan State University)  相似文献   

17.
目的 探讨非栓塞性脑梗死患者发生出血性转化(hemorrhagic transformation,HT)的病因、临床特点及治疗.方法 回顾性分析19例非栓塞性脑梗死患者的临床资料.结果 患者平均年龄为76.4岁,大面积脑梗死(95%)是主要病因,高血压(74%)、糖尿病(53%)亦有较高的发生率.结论 非栓塞性脑梗死的HT常发生在高龄及大面积脑梗死,高血压、糖尿病亦是重要的病因,多为晚发型出血,预后相对较好.
Abstract:
Objective To investigate the etiological factors, clinical characteristics and cures of patients with nonembolic cerebral infarction after cerebral infarction. Methods The clinical data of 19 patients with non-embolic cerebral infarction were analyzed retrospectively. Results The average age of patients was 76.4 and the major risk factor was massive cerebral infarction (95%). While hypertension (74%) and diabetes mellitus (53%) were also the risk factors of nonembolic cerebral infarction. Conclusions Hemorrhagic transformation after non-embolic cerebral infarction usually occurs in patients with old age and massive cerebral infarction. Hypertension and diabetes mellitus are also the major risk factors for hemorrhagic transformation after non-embolic cerebral infarction, mainly with delayed onset and better outcomes.  相似文献   

设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号