Plotting column values per group, but with a color based on another column

Question

With a dataframe structured loosely in the way shown below.

                                       ID  Price  ColorID
Datetime                                                 
2020-10-30 07:21:00.074000+00:00   131310  13.13        0
2020-10-30 07:21:00.485128+00:00  5250455  13.02        1
2020-10-30 07:21:00.485000+00:00   130210  13.02        0
2020-10-30 07:21:00.486000+00:00   130110  13.01        0
2020-10-30 07:21:00.486000+00:00   130310  13.03        0
2020-10-30 07:21:00.488000+00:00   130310  13.03        0

I'm trying to draw price progression per ID, with the colorID defining the color to plot in.

If I don't use any colors, the below command will give me the plot I am looking for, but with an inbuilt color scheme, rather than the one I'm requiring..

test.groupby("ID").Price.plot(drawstyle="steps-post", marker="d", linewidth=2)

Same ID will retain same colorID, so first trial has been to first create a grouped object, then refer to that grouped object to try and obtain a color scheme.

grouped = test.groupby("ID")
colors = {0: "b", 1: "r"}
grouped.Price.plot(drawstyle="steps-post", marker="d", linewidth=2, c=colors[grouped.ColorID.last()])

which runs into the error:

TypeError: 'Series' objects are mutable, thus they cannot be hashed

whereas trying to go after a direct application through

grouped.Price.plot(drawstyle="steps-post", marker="d", linewidth=2,c=np.where(grouped.ColorID==0,"b","r"))

will try to read the color array as RGB channel, hence resulting in

ValueError: Invalid RGBA argument: array('r', dtype='<U1')

Is there a straightforward way I'm blatantly missing, or am I consigned to either making multi-label groups or plotting 1 by 1 in a for loop, specifying the plot handles outside the loop?

Quang Hoang · Accepted Answer · 2020-11-10 16:43:55Z

1

I know you are hesitant to for loop, but it's not really a bad thing. It provides you more control on how the plot looks. Plus, I doubt that you get any vectorization with groupby().plot().

So a for loop would look like:

colors = {0: "b", 1: "r"}

grouped = test.groupby("ID")

fig, ax = plt.subplots()
for k, g in grouped:
    color = colors[g['ColorID'].iloc[0]]
    g.Price.plot(drawstyle="steps-post", marker="d", 
                 linewidth=2, c=color,
                 label=k, ax=ax)

ax.legend()

Output:

answered Nov 10, 2020 at 16:43

Quang Hoang

151k11 gold badges64 silver badges86 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

Uvar Over a year ago

You are probably right that internally a for loop is used anyway, and I should just use it explicitly under my own control rather than under the hood. Thanks.

Uvar Over a year ago

One issue with doing it in the subplots way, however, is that I can't really make a hovering annotation for if I wish to check in-plot which ID I'm looking at.

Quang Hoang Over a year ago

@Uvar That's just my usual way to plot. You can skip all subplots related code.

Collectives™ on Stack Overflow

Plotting column values per group, but with a color based on another column

1 Answer 1

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related