我有一个包含32个变量和48个观察值的数据集（观察值会增加，因为我们正处于通过订阅的数据收集阶段）。 由于数据集具有较高的多重共线性，因此我介绍了Farrar – Glauber检验。测试的单个度量（成语）具有一个称为Klein的参数，该参数的值分别为0s和1s，表示变量是否具有多重共线性。 现在基于Klien的值，我需要从主数据集中删除这些列。我被困在如何基于数据集imcdiag_idiagval中的Klein值从主数据集（model_df）中删除/删除。
我的以下示例代码， model_df0：不包含目标变量的数据集 model_df：具有目标变量的数据集
library(mctest)
#Farrar – Glauber test(F–test) for location of multicollinearity
# Individual Multicollinearity Diagnostic Measures
imcdiag_out = imcdiag(model_df0, model_df$TARGET)
summary(imcdiag_out)
#Output of summary function below,
Length Class Mode
idiags 224 -none- numeric
x 1536 -none- numeric
y 48 -none- numeric
method 0 -none- NULL
corr 1 -none- logical
call 3 -none- call
pval 29 -none- numeric
R2 1 -none- numeric
all 1 -none- logical
alldiag 224 -none- logical
imcdiag_idiagval = as.data.frame(imcdiag_out$idiags)
> imcdiag_idiagval$Klein
[1] 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0
在32个变量中，Klein中值为1的值具有多重共线性。因此需要删除它们
> dput(head(model_df))
structure(list(Mon = structure(c(2014.33333333333, 2014.41666666667,
2014.5, 2014.58333333333, 2014.66666666667, 2014.75), class = "yearmon"),
Al_Prod_Qty = c(4516084.63333333, 4488436.58064516, 4529767.5,
4468792.16129032, 4515652.4516129, 4513669.13333333), US_Indus_Growth = c(2.72933333333333,
3.00806451612903, 3.25466666666667, 3.41161290322581, 4.03322580645161,
3.22333333333333), China_Elec_Rail = c(6.206, 7.30733333333333,
7.18741935483871, 5.84322580645161, 3.90166666666667, 5.65290322580645
), DS_Index = c(80.4, 79.81, 81.52, 82.78, 86.05, 87.02),
Al_Prod_Africa = c(152L, 143L, 144L, 148L, 142L, 144L), Al_Prod_NorthAmerica = c(392L,
372L, 386L, 386L, 372L, 379L), Al_Prod_SouthAmerica = c(135L,
117L, 117L, 116L, 111L, 120L), Al_Prod_AsiaEXChina = c(203L,
193L, 203L, 198L, 194L, 204L), Al_Prod_WestEurope = c(304L,
297L, 306L, 306L, 298L, 307L), Al_Prod_EastCentralEurope = c(319L,
309L, 319L, 319L, 310L, 321L), Al_Prod_Oceania = c(178L,
174L, 175L, 166L, 160L, 165L), Al_Prod_GCC = c(418L, 412L,
427L, 426L, 413L, 426L), Al_Prod_ChinaEstimated = c(12, 15,
18, 22, 23, 24), Al_Prod_Total = c(4441L, 4419L, 4507L, 4553L,
4502L, 4619L), Al_Prod_DailyAvg = c(143.3, 147.3, 145.4,
146.9, 150.1, 149), almna_prod_AfricaAsiaExChina = c(395L,
445L, 459L, 436L, 413L, 395L), almna_prod_NorthAmerica = c(440L,
444L, 479L, 476L, 441L, 455L), almna_prod_SouthAmerica = c(1137L,
1077L, 1047L, 1063L, 1068L, 1097L), almna_prod_WestEurope = c(326L,
363L, 334L, 386L, 351L, 323L), almna_prod_EastCentralEurope = c(352L,
322L, 330L, 321L, 325L, 325L), almna_prod_Oceania = c(1696L,
1632L, 1663L, 1631L, 1579L, 1693L), almna_prod_ChinaEstimated = c(4513L,
4448L, 4509L, 4461L, 4416L, 4754L), almna_prod_Total = c(8859L,
8731L, 8821L, 8774L, 8593L, 9042L), almna_prod_DailyAverage = c(285.77,
291.03, 284.55, 283.03, 286.43, 291.68), TARGET = c(1749.1,
1834.4, 1945.41, 2030.23, 1992.48, 1938.2), lme_al_3month_price = c(1790.98,
1868.83, 1967.28, 2038.1, 2023.48, 1956.39), lme_al_stock = c(5271645,
5116815, 4990208, 4887148, 4706865, 4519988), Coal_Central_Appalachia_Price = c(60.7,
60.05, 57.25, 56.35, 54.59, 56.3), Coal_Northern_Apppalachia_Price = c(65.5,
64.5, 63.7, 62.75, 63.3, 65.55), Coal_Illinois_Basin_Price = c(46,
45.45, 44, 44, 43.4, 44.7), Coal_Powder_River_Price = c(13,
12.55, 12.25, 11.55, 11.25, 11.55), Coal_Uinta_Basin_Price = c(36.75,
36.75, 36.45, 36.45, 37, 37.75), BrentCrudePrice = c(109.21,
111.03, 104.94, 101.12, 94.67, 84.17)), class = c("data.table",
"data.frame"), row.names = c(NA, -6L), .internal.selfref = <pointer: 0x00000000025f1ef0>)
答案 0 :(得分：1)
如果1表示变量不是多共线性，则流代码：
var query = "SELECT * FROM conversation WHERE participants CONTAINS '1stvalue' AND participants CONTAINS '2ndvalue' ALLOW FILTERING;";
models.instance.Conversation.execute_query(query, {}, function(err, Conversations){
//Conversation is an array of plain objects
});
您的数据中有34列：
#raw data
dataset<-sapply(paste0("var",1:32), function(x) assign(x,rnorm(n=48)) )
imcdiag_idiagval.Klein<-c(0 ,0 ,0, 0 ,0 ,0, 0 ,0, 0 ,0 ,0 ,0 ,0, 1, 1, 1, 1, 1, 1 ,1 ,1, 1, 1 ,1 ,0 ,0, 0, 0, 0, 0, 0, 0)
#final data without multi-collinearity
final_dataset<-dataset[,imcdiag_idiagval.Klein==1]
所以您需要一个33的向量：
names(data)
[1] "Mon" "Al_Prod_Qty"
[3] "US_Indus_Growth" "China_Elec_Rail"
[5] "DS_Index" "Al_Prod_Africa"
[7] "Al_Prod_NorthAmerica" "Al_Prod_SouthAmerica"
[9] "Al_Prod_AsiaEXChina" "Al_Prod_WestEurope"
[11] "Al_Prod_EastCentralEurope" "Al_Prod_Oceania"
[13] "Al_Prod_GCC" "Al_Prod_ChinaEstimated"
[15] "Al_Prod_Total" "Al_Prod_DailyAvg"
[17] "almna_prod_AfricaAsiaExChina" "almna_prod_NorthAmerica"
[19] "almna_prod_SouthAmerica" "almna_prod_WestEurope"
[21] "almna_prod_EastCentralEurope" "almna_prod_Oceania"
[23] "almna_prod_ChinaEstimated" "almna_prod_Total"
[25] "almna_prod_DailyAverage" "TARGET"
[27] "lme_al_3month_price" "lme_al_stock"
[29] "Coal_Central_Appalachia_Price" "Coal_Northern_Apppalachia_Price"
[31] "Coal_Illinois_Basin_Price" "Coal_Powder_River_Price"
[33] "Coal_Uinta_Basin_Price" "BrentCrudePrice"
添加响应变量：
imcdiag_idiagval.Klein<-sample(0:1,33,replace = T)
最终数据：
fina_index<-c(imcdiag_idiagval.Klein,1)
比较：
final_data<-data[,fina_index==1]
答案 1 :(得分：0)
这里是一个示例，其中1表示应使用mtcars示例数据从数据帧中删除一列。
data(mtcars)
# set up true false variable where 1 = drop, 0 = keep
# noting that it is of type numeric(), not logical()
drop <- c(0,0,0,0,1,1,1,1,0,0,0)
# this will result in a data frame that
# contains 7 columns, excluding drat, wt, qsec, and vs
theSubset <- mtcars[!drop]
head(theSubset)
...以及输出：
> head(theSubset)
mpg cyl disp hp am gear carb
Mazda RX4 21.0 6 160 110 1 4 4
Mazda RX4 Wag 21.0 6 160 110 1 4 4
Datsun 710 22.8 4 108 93 1 4 1
Hornet 4 Drive 21.4 6 258 110 0 3 1
Hornet Sportabout 18.7 8 360 175 0 3 2
Valiant 18.1 6 225 105 0 3 1
>