在R中删除索引/值后如何保留它们

Question

是否可以在删除值和索引后保留它们？ 我的意思是我很想测试某些插值方法，我想删除一些值，运行插值，然后将创建的新值与我之前删除的值进行比较。因此，我不会出现随机删除某些值的问题，但是我不知道如何继续保留它们。我希望我的问题很清楚，

最诚挚的问候

马克西姆

Answer 1

我认为，对该问题的最佳答案是挑战其前提的一个答案：您不必删除插值即可进行另一种形式的插值。一个例子：

xvals <- 1:10 # These are the observed x values
yvals <- sin(xvals) # The observed y values
linear_interpolation <- approx(xvals, yvals) # linear interpolation of values
spline_interpolation <- spline(xvals, yvals, n = 50) # spline interpolation
# We can collect the interpolation results in a dataframe for easy comparison
comparison <- data.frame(x = linear_interpolation$x,
                         linear_interpolation = linear_interpolation$y,
                         spline_interpolation = spline_interpolation$y)
head(comparison) # Look at the first few rows
#>          x linear_interpolation spline_interpolation
#> 1 1.000000            0.8414710            0.8414710
#> 2 1.183673            0.8539289            0.9476460
#> 3 1.367347            0.8663868            1.0066238
#> 4 1.551020            0.8788447            1.0227805
#> 5 1.734694            0.8913027            1.0004924
#> 6 1.918367            0.9037606            0.9441358
# We can also plot them to compare visually
# I like to use the colorblind-friendly palette from
# Wong, Bang. 2011. "Points of view: Color blindness." Nature Methods 8:441.
wong_palette <- c("#e69f00", "#56b4e9", "#009e73")
# Start with the observed values
plot(xvals, yvals, pch = 19, xlab = "X Values", ylab = "Y Values",
     main = "Interpolation Comparison")
# Plot the linear interpolation
lines(linear_interpolation, col = wong_palette[1])
# The spline interpolation
lines(spline_interpolation, col = wong_palette[2])
# And, since we know the true function in this simulation, the true function
lines(linear_interpolation$x, sin(linear_interpolation$x), col = wong_palette[3])
# Finish with a legend so we know what we're looking at
legend("bottomleft", seg.len = 1, lty = 1, bty = "n", col = wong_palette,
       legend = c("Linear Interp.", "Spline Interp.", "True Function"))

^{由reprex package（v0.2.1）于2018-11-18创建}

Answer 2

最后我确实喜欢这个（不是很传统，但是..）：

创建数据框

data_test<-data.frame( x<-1:10, y<-1:10 ) data_test

通过NA随机替换某些值

data_test$y [sample(nrow(data_test),2)]<- NA data_test

使用那些NA查找索引

index<-which(data_test$y %in% c(NA)) index

保留这些已删除的值

store_data<-data_test[index,]

在df中使用NA删除值

data_testnew<- na.omit(data_test) data_testnew