Series Hack

Sometimes you have several sources of data, and you want to visualize them by displaying color mapping from a group (i.e., color of plot objects corresponds to their group).

Let's assume we have three samples with different y-values and the same x-values

val xs = listOf(1, 2, 3, 4, 5)
val ysA = listOf(1.0, 2.5, 3.0, 3.5, 5.0)
val ysB = listOf(0.5, 1.5, 3.0, 1.5, 0.0)
val ysC = listOf(3.0, 5.0, 2.0, 3.0, 5.0)

So you can try something like that:

plot {
    x(xs)
    line {
        y(ysA)
        color = Color.RED
    }
    line {
        y(ysB)
        color = Color.GREEN
    }
    line {
        y(ysC)
        color = Color.BLUE
    }
}

But unfortunately, "lets-plot" doesn't support it — legend is not displayed (because of Lets-Plot limitations). So you need to gather them into one layer. You need to change your data - add grouping variable (so now each point has exactly three parameters: x and y coordinates and its group).

plot {
    line {
        x(xs + xs + xs)
        y(ysA + ysB + ysC)
        color(xs.map { "A" } + xs.map { "B" } + xs.map { "C" })
    }
}

DataFrame solution

In case a similar situation occurred with a dataframe, it is even easier to solve it with the gather function

val df = dataFrameOf(
    "xs" to xs,
    "ysA" to ysA,
    "ysB" to ysB,
    "ysC" to ysC,
)
df

xs	ysA	ysB	ysC
1	1	0.5	3
2	2.5	1.5	5
3	3	3	2
4	3.5	1.5	3
5	5	0	5

For example, we cant to build several bars:

// Doesn't work - bars are overlapped, and we can't change that
df.plot {
    x(xs)
    bars {
        y(ysA)
        fillColor = Color.RED
    }
    bars {
        y(ysB)
        fillColor = Color.GREEN
    }
    bars {
        y(ysC)
        fillColor = Color.BLUE
    }
}

val gatheredDF = df.gather {
    ysA and ysB and ysC
}
    .mapKeys { it.drop(2) } // Take group name as a key
    .into(keyColumn = "group", valueColumn = "ys")
gatheredDF.head()

xs	group	ys
1	A	1
1	B	0.5
1	C	3
2	A	2.5
2	B	1.5

gatheredDF.plot {
    bars {
        x(xs)
        y(ys)
        fillColor(group)
    }
}

Last modified: 07 December 2023