Question

我想在R中制作交互图，如示例

http://www.ling.upenn.edu/~joseff/rstudy/week4.html#interactionplot

除了代替我想要通过除数字之外的某些属性为点着色的线。所需的功能看起来像

blood_pressure = ...  # this will be the x-axis
age_bucket = ... # this will be the interaction-level
gender = ... # color boys blue, girls red
stroke_rate = ... # y-axis

interaction.plot(blood_pressure, age_bucket, stroke_rate, col=gender, type="b"...

除了color =性别实际上做了我想要的

Answer 1

问题在于interaction.plot实际上依赖于tapply()来计算汇总的汇总度量（平均值或中位数）。由于interaction.plot只是调用matplot，您可以根据需要直接使用后者，或者使用plyr汇总数据，并使用ggplot2绘制结果（更灵活）。< / p>

# consider the 64x4 dataset OrchardSprays and create a fake
# two-levels factor, say grp, which is in correspondence to rowpos odd/even values
grp <- gl(2, 1, 8, labels=letters[1:2])
# the following won't work (due to color recycling)
with(OrchardSprays, 
     interaction.plot(treatment, rowpos, decrease, 
                      type="b", pch=19, lty=1, col=as.numeric(colpos), legend=F))
# what is used to draw the 8 lines is computed from
with(OrchardSprays, 
     tapply(decrease, list(treatment=treatment, rowpos=rowpos), mean))
# the following will work, though
with(OrchardSprays, 
     interaction.plot(treatment, rowpos, decrease, 
                      type="b", pch=19, lty=1, col=as.numeric(grp), legend=F))

简而言之，假设您可以在gender与跟踪因子（age_bucket）之间找到足够的映射，则需要构建大小为nlevels(age_bucket)的颜色矢量。

在interaction.plot（R）I中着色点而不是线

1 个答案: