CODEWITHSUNDEEP
pythonaltair
djcunningham0
djcunningham0

Reputation: 187

Plotting two series using different colors and marks in Altair

I have a Pandas dataframe that looks like this:

| date_column         |   my_point_column |   my_line_column |
|---------------------|-------------------|------------------|
| 2017-12-13 00:00:00 |                 3 |          9.73136 |
| 2018-01-05 00:00:00 |                23 |          6.36568 |
| 2018-01-11 00:00:00 |                 6 |          11.9105 |
| 2018-01-22 00:00:00 |                11 |          10.4328 |
| ...                 |               ... |              ... |

I want to plot both series using different marks (points and lines) and they should each be a different color. The color also must be shown in the legend.

I was able to accomplish it with this code, but it's pretty verbose:

# plot_df is the dataframe shown above
chart = alt.Chart(plot_df)

points = chart.transform_fold(
    fold=["my_point_column"],
    as_=["variable", "value"]
).encode(
    x="date_column:T",
    y="my_point_column:Q",
    color="variable:N"
).mark_circle()

line = chart.transform_fold(
    fold=["my_line_column"],
    as_=["variable", "value"]
).encode(
    x="date_column:T",
    y="my_line_column:Q",
    color="variable:N"
).mark_line()

alt.layer(points, line)

That code produces this chart:

point and line chart

That's pretty much the result I'm looking for (outside of customizing the colors, axis titles, etc.). Is there an easier way to do it without having to use transform_fold twice?

Upvotes: 1

Views: 1362

Answers (1)

joelostblom
joelostblom

Reputation: 48879

For a less verbose solution, you could use one of the charts as the "base" and then overwrite its transform, mark, and y-encoding:

points = alt.Chart(plot_df).mark_circle().transform_fold(
    fold=["my_point_column"]
).encode(
    x="date_column:T",
    y="my_point_column:Q",
    color="key:N")

line = chart.mark_line().transform_fold(
    fold=["my_line_column"]
).encode(
    y="my_line_column:Q")

points + line

Upvotes: 1

Related Questions