Question

我最近从Stata迁移到R.我不确定如何对分组和未分组的观察进行计算描述性统计。

这是我的数据：

dput(DF)
structure(list(Product_Name = c("iPhone", "iPhone", "iPhone", 
"iPhone", "iPhone", "iPhone", "Nexus 6P", "Nexus 6P", "Nexus 6P", 
"Nexus 6P", "Nexus 6P", "Nexus 6P"), Product_Type = c("New", 
"New", "Refurbished", "New", "New", "Refurbished", "Refurbished", 
"Refurbished", "Refurbished", "Refurbished", "Refurbished", "Refurbished"
), Year = c(2006, 2011, 2009, 2008, 2011, 2009, 2012, 2007, 2013, 
2015, 2009, 2010), Units = c(100, 200, 300, 400, 500, 600, 700, 
200, 120, 125, 345, 340)), .Names = c("Product_Name", "Product_Type", 
"Year", "Units"), row.names = c(NA, 12L), class = "data.frame")

我的数据包含按年份和类型销售的产品。每种产品都可以是翻新产品或新产品。此外，如果它在2010年之前出售，我会将其标记为在＃34;时间1＆＃34;否则我会把它标记为＆＃34;时间2＆＃34;。

这是我的代码：

DF[DF$Year<2010,"Time"]<-"1"
DF[DF$Year>=2010,"Time"]<-"2"

现在，我想为这些时间段生成描述性统计数据。

DF %>% 
  group_by(Product_Name, Product_Type,Time) %>%
  dplyr::summarise(Count = n(), 
                   Sum_Units = sum(Units,na.rm=TRUE), 
                   Avg_Units = mean(Units,na.rm = TRUE), 
                   Max_Units=max(Units,na.rm = TRUE))

如果我们运行上面的代码，我们会按Product_Name，Product_Type和Time获取描述性统计信息（即分组描述性统计信息）。但是，这不是我想要的。我想要使用和不考虑Product_Type和Time的分组进行描述性统计。意思是，我想计算描述性统计数据，假设产品在时间1或时间2（即所有年份）出售，而不管销售的产品类型如何，同时保留上面的一些分组信息。

预期输出：

dput(DFOut)
structure(list(Product_Name = c("iPhone", "Nexus 6P"), New_Units_Sum_Time1 = c(500, 
NA), Refurbished_Units_Sum_Time_1 = c(900, 545), Sum_Units_Time1 = c(1400, 
545), Sum_Units_Time2 = c(700, 1285), Sum_Units_Time_1_And_2 = c(2100, 
1830), Avg_Units_Time1 = c(350, 272.5), Avg_Units_Time2 = c(350, 
321.25), Avg_Units_Time_1_And_2 = c(350, 305), Max_Units_Time1 = c(600, 
345), Max_Units_Time2 = c(500, 700), Max_Units_Time_1_And_2 = c(600, 
700)), .Names = c("Product_Name", "New_Units_Sum_Time1", "Refurbished_Units_Sum_Time_1", 
"Sum_Units_Time1", "Sum_Units_Time2", "Sum_Units_Time_1_And_2", 
"Avg_Units_Time1", "Avg_Units_Time2", "Avg_Units_Time_1_And_2", 
"Max_Units_Time1", "Max_Units_Time2", "Max_Units_Time_1_And_2"
), row.names = 1:2, class = "data.frame")

在输出中，您会看到我有一些描述性统计信息：

a）基于产品类型及其销售周期（例如New_Units_Sum_Time1，即New和Time1）。请注意，在输出中，我只显示了New和Time1的组合。如果您可以指导我如何为Refurbished和Time的其他组合生成描述性统计信息，那就太棒了。

b）基于忽略产品类型但不忽略销售期限（例如Sum_Units_Time1，即Time1）

c）基于忽略产品类型和销售期限（例如Sum_Units_Time_1_And_2）。

同上Avg和Mean。

我该怎么做？我很感激任何帮助。我真的很挣扎。

请注意，我使用Excel手动创建了DFOut。虽然我对其进行了三重检查，但可能存在一些手动错误 - 如果有问题，我会非常乐意澄清它们。谢谢你的时间。

sessionInfo（）

R version 3.3.2 (2016-10-31)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows >= 8 x64 (build 9200)

locale:
[1] LC_COLLATE=English_United States.1252  LC_CTYPE=English_United States.1252   
[3] LC_MONETARY=English_United States.1252 LC_NUMERIC=C                          
[5] LC_TIME=English_United States.1252    

attached base packages:
[1] grDevices datasets  stats     graphics  grid      tcltk     utils     methods   base     

other attached packages:
 [1] tables_0.8              Hmisc_4.0-2             Formula_1.2-1           survival_2.40-1        
 [5] ResourceSelection_0.3-0 magrittr_1.5            stringr_1.1.0           bit64_0.9-5            
 [9] bit_1.1-12              tufterhandout_1.2.1     knitr_1.15.1            rmarkdown_1.3          
[13] tufte_0.2               corrplot_0.77           purrr_0.2.2             readr_1.0.0            
[17] tibble_1.2              tidyverse_1.1.1         cowplot_0.7.0           plotly_4.5.6           
[21] ggplot2_2.2.1           maps_3.1.1              directlabels_2015.12.16 tidyr_0.6.1            
[25] ggthemes_3.3.0          R2HTML_2.3.2            lubridate_1.6.0         xts_0.9-7              
[29] zoo_1.7-14              lattice_0.20-34         corrgram_1.10           hexbin_1.27.1          
[33] sm_2.2-5.4              compare_0.2-6           installr_0.18.0         psych_1.6.12           
[37] reshape2_1.4.2          readstata13_0.8.5       pastecs_1.3-18          boot_1.3-18            
[41] vcd_1.4-3               car_2.1-4               xlsxjars_0.6.1          rJava_0.9-8            
[45] debug_1.3.1             dplyr_0.5.0             foreign_0.8-67          gmodels_2.16.2         
[49] openxlsx_4.0.0          plyr_1.8.4             

loaded via a namespace (and not attached):
 [1] minqa_1.2.4         colorspace_1.3-2    class_7.3-14        modeltools_0.2-21   mclust_5.2.2       
 [6] rprojroot_1.2       htmlTable_1.9       base64enc_0.1-3     MatrixModels_0.4-1  flexmix_2.3-13     
[11] mvtnorm_1.0-5       xml2_1.1.1          codetools_0.2-15    splines_3.3.2       mnormt_1.5-5       
[16] robustbase_0.92-7   jsonlite_1.2        nloptr_1.0.4        pbkrtest_0.4-6      broom_0.4.1        
[21] cluster_2.0.5       kernlab_0.9-25      httr_1.2.1          backports_1.0.5     assertthat_0.1     
[26] Matrix_1.2-7.1      lazyeval_0.2.0      acepack_1.4.1       htmltools_0.3.5     quantreg_5.29      
[31] tools_3.3.2         gtable_0.2.0        Rcpp_0.12.9         trimcluster_0.1-2   gdata_2.17.0       
[36] nlme_3.1-128        iterators_1.0.8     fpc_2.1-10          lmtest_0.9-34       lme4_1.1-12        
[41] rvest_0.3.2         gtools_3.5.0        dendextend_1.4.0    DEoptimR_1.0-8      MASS_7.3-45        
[46] scales_0.4.1        TSP_1.1-4           hms_0.3             parallel_3.3.2      SparseM_1.74       
[51] RColorBrewer_1.1-2  gridExtra_2.2.1     rpart_4.1-10        latticeExtra_0.6-28 stringi_1.1.2      
[56] gclus_1.3.1         mvbutils_2.7.4.1    foreach_1.4.3       checkmate_1.8.2     seriation_1.2-1    
[61] caTools_1.17.1      prabclus_2.2-6      bitops_1.0-6        evaluate_0.10       htmlwidgets_0.8    
[66] R6_2.2.0            gplots_3.0.1        DBI_0.5-1           haven_1.0.0         whisker_0.3-2      
[71] mgcv_1.8-16         nnet_7.3-12         modelr_0.1.0        KernSmooth_2.23-15  viridis_0.3.4      
[76] readxl_0.1.1        data.table_1.10.0   forcats_0.2.0       digest_0.6.12       diptest_0.75-7     
[81] stats4_3.3.2        munsell_0.4.3       registry_0.3        viridisLite_0.1.3   quadprog_1.5-5

Answer 1

自动执行此操作的一种方法是首先创建一个向量（ind），其中包含所有可能的分组变量组合。然后，我们将这些组合转换为Units的公式。由于每个公式都保存在列表（l1）中，我们迭代该列表并进行汇总。

ind <- unlist(sapply(c(2,3), function(i) combn(c('Product_Name', 'Product_Type', 'Time'), 
                                                             i, paste, collapse = '+')))

l1 <- sapply(ind, function(i) as.formula(paste('Units ~ ', i)))

lapply(l1, function(i) aggregate(i, df, FUN = function(j) c(sum1 = sum(j), 
                                                            avg = mean(j), 
                                                            max_units = max(j))))

#which gives

#$`Product_Name+Product_Type`
#  Product_Name Product_Type Units.sum1 Units.avg Units.max_units
#1       iPhone          New       1200       300             500
#2       iPhone  Refurbished        900       450             600
#3     Nexus 6P  Refurbished       1830       305             700

#$`Product_Name+Time`
#  Product_Name Time Units.sum1 Units.avg Units.max_units
#1       iPhone    1    1400.00    350.00          600.00
#2     Nexus 6P    1     545.00    272.50          345.00
#3       iPhone    2     700.00    350.00          500.00
#4     Nexus 6P    2    1285.00    321.25          700.00

#$`Product_Type+Time`
#  Product_Type Time Units.sum1 Units.avg Units.max_units
#1          New    1     500.00    250.00          400.00
#2  Refurbished    1    1445.00    361.25          600.00
#3          New    2     700.00    350.00          500.00
#4  Refurbished    2    1285.00    321.25          700.00

#$`Product_Name+Product_Type+Time`
#  Product_Name Product_Type Time Units.sum1 Units.avg Units.max_units
#1       iPhone          New    1     500.00    250.00          400.00
#2       iPhone  Refurbished    1     900.00    450.00          600.00
#3     Nexus 6P  Refurbished    1     545.00    272.50          345.00
#4       iPhone          New    2     700.00    350.00          500.00
#5     Nexus 6P  Refurbished    2    1285.00    321.25          700.00

关于分组和未分组观测的描述性统计

1 个答案: