wideLM(WideLM)
wideLM()所属R语言包：WideLM
Fitting Multiple Models of Modest Size
配件多种型号的适度规模
译者：生物统计家园网 机器人LoveR
描述----------Description----------
Fits numerous linear models to a fixed design and response. The predictors are specified via a list of parallel index vectors and the interaction terms are computed on the fly. The design matrix is currently restricted to contain 1024 rows or less. A vector of indices into the columns of the response matrix is used to specify the response to be fit.
适用于众多的线性模型到一个固定的设计和响应。通过列表的并行索引向量指定的预测变量的交互项的计算上飞。目前仅限于包含1024行或更小的设计矩阵。响应矩阵的列的索引的是，一种向量，用于指定响应定为适宜。
用法----------Usage----------
wideLM(x, y, yIdx, preds, formulaMode, addendum=NULL)
参数----------Arguments----------
参数：x
the design matrix.
设计矩阵。
参数：y
a matrix of responses. The responses are organized as columns. Hence x and y should have the same number of rows.
响应的矩阵。响应组织列。因此，x和y应该有相同的行数。
参数：yIdx
a vector of column indices into the response matrix
到响应矩阵列索引的向量
参数：preds
A list of vectors of column indices into the design matrix. Together, these vectors specify the predictors to be fit in each individual model. All vectors should have the same length, that is, the total number of models to be tested. N.B.: The package is currently limited to two predictors.
列索引的矢量设计矩阵列表。总之，这些向量指定适合每个模型的预测。所有向量应具有相同的长度，也就是，以进行测试的模型的总数。注：包目前仅限于两个预测。
参数：formulaMode
"additive", "saturated", "accumulated", "reduced". The terms additive and saturated refer to types of models. The terms accumulated and reduced are shorthand for additions to the additive and deletions from the saturated model, respectively, using the addendum mechanism described below.
“添加剂”，“饱和”，“累积”，“减少”。的条款additive和saturated类型的模型。的条款accumulated和reduced是速记的additive和删除从saturated模型，分别使用addendum机制。
参数：addendum
a vector of exceptions to the additive model or saturated model. Exceptions give a compact way to describe models somewhat larger than the additive model or somewhat smaller than the saturated model. Exceptions are integers which enumerate the possible interaction terms in a saturated model on 'n' predictors. Hence their values range from 1 to 2^n - (n+1). Exceptions only apply to the accumulated and reduced modes. For the accumulated mode, all enumerated interaction terms are appended to the additive model. For the reduced mode, the enumerated interaction terms are deleted from the saturated model. The actual mapping is based on the lexical ordering of subsets of a set of size 'n'. This is explained intuitively via following schema, for predictors 'p_1', 'p_2', ..., 'p_n': The numbers from 1 to (n-choose-two) name interaction pairs: 1, ..., n-1 refer to 'p_1 * p_2', ... ,'p_1 * p_n', respectively. The next 'n-2' numbers refer to 'p_2 * p_3' , ..., 'p_2 * p_n' ... and so forth until all pairs have been enumerated. The next (n-choose-three) numbers refer to triples: 'p_1 * p_2 * p_3', 'p_1 * p_2 * p_4', ..., 'p_1 * p_2 * p_n' 'p_2 * p_3 * p_4', ..., 'p_2 * p_3 * p_n' ... and so forth until all triples have been enumerated. This scheme is repeated until the single n-fold product is enumerated, which should have value 2^n - (n+1).
一个向量的加法模型的例外情况或饱和模型。异常给一个紧凑的方式描述稍大比添加剂模型或稍微小于饱和模型的模型。例外的是整数，枚举可能的相互作用在饱和模式N的预测。因此，它们的取值范围从1到2 ^ N - （N +1）。例外accumulated和reduced模式只适用于。对于accumulated模式，所有列举的互动方面添加的添加剂模型中。 reduced模式，所列举的交互项被删除饱和模型。实际映射是根据大小“N”的一组的子集上的词汇顺序。这是解释直观地通过下面的架构，为的预测“P_1”，“P_2，...，”P_N：（N - 请选择 - 2）名称相互作用对数字从1到1，...，N- 1参考“P_1 * P_2”，... ，P_1 * P_N，分别。下一个n-2个的数字指的P_2 * P_3，...，P_2 * P_N...依此类推，直到所有对已列举了。接下来的第（n-选择在三个）的数字指的三元组：P_1 * P_2 * P_3，P_1 * P_2 * P_4，...，P_1 * P_2 * P_NP_2 * P_3 * P_4。 ...“，P_2 * P_3 * P_N”，...依此类推，直到所有的三元组已列举了。该方案反复进行，直至被列举的n倍的产品，它应该有值2 ^ N - （N +1）。
值----------Value----------
A list containing the matrices coef of coefficients and tscore of t-scores. The columns refer to the indices of the vectors given in the preds list. The rows are ordered as follows:
一个列表，其中包含矩阵coef系数和tscoreT值。列指preds列表中给出的向量的索引。的行进行排序，如下所示：
intercept, predictors, interaction terms (in lexical order).
拦截，预测，交互项（词法顺序）。
（作者）----------Author(s)----------
Mark Seligman at Insilicos LLC. Original Fortran implementation
computing the t-scores was developed by Chris Fraley at Insilicos LLC.
转载请注明:出自 生物统计家园网(http://www.biostatistic.net)。
注：
注1：为了方便大家学习，本文档为生物统计家园网机器人LoveR翻译而成，仅供个人R语言学习参考使用，生物统计家园保留版权。
注2：由于是机器人自动翻译，难免有不准确之处，使用时仔细对照中、英文内容进行反复理解，可以帮助R语言的学习。
注3：如遇到不准确之处，请在本贴的后面进行回帖，我们会逐渐进行修订。
|
发表于 2012-10-1 22:28:45
