如何检测和绘制变化？

Question

在示例数据框中：

dframe <- structure(list(id = c(14768L, 18180L), col1 = c(-0.6084, -0.3227
), col2 = c(-1.4887, -1.1797), col3 = c(3.8402, 3.0491), col4 = c(-1.8265, 
-1.3248), col5 = c(0.4078, 0.7862), col1_new = c(-0.4582, -0.2094
), col2_new = c(-1.3878, -1.5926), col3_new = c(3.3112, 3.2756
), col4_new = c(-1.6242, -1.2361), col5_new = c(0.5014, 0.5925
)), class = "data.frame", row.names = c(NA, -2L))

如何检测所有唯一的id的唯一ID，即每个对col名称（即col1-col1_new）的更改是否增加或减少

Answer 1

首先，使用data.table::melt转换为更有用的长格式会有所帮助。然后，我们可以简单地计算出差异，并添加一个新列以指示增加或减少。

library(data.table)
library(tidyverse)

setDT(dframe)

# take each pair of original and new values, and move each pair from columns 
# to their own row (with two columns, "current" and "new"
dframe_new <- melt(dframe, id.var = "id", measure = patterns("\\d$", "new$"), 
    value.name = c("current", "new")) %>% 
mutate(
    diff = new-current, 
    Change = case_when(
        diff > 0 ~ "Increase",
        diff == 0 ~ "No Change",
        TRUE ~ "Decrease"
    )
)

      id variable current     new    diff   Change
1  14768        1 -0.6084 -0.4582  0.1502 Increase
2  18180        1 -0.3227 -0.2094  0.1133 Increase
3  14768        2 -1.4887 -1.3878  0.1009 Increase
4  18180        2 -1.1797 -1.5926 -0.4129 Decrease
5  14768        3  3.8402  3.3112 -0.5290 Decrease
6  18180        3  3.0491  3.2756  0.2265 Increase
7  14768        4 -1.8265 -1.6242  0.2023 Increase
8  18180        4 -1.3248 -1.2361  0.0887 Increase
9  14768        5  0.4078  0.5014  0.0936 Increase
10 18180        5  0.7862  0.5925 -0.1937 Decrease

一个简单的情节可能是：

ggplot(dframe_new, aes(factor(id), diff, color = Change)) + geom_point()