How facet resolves multiple records for the facetted parameter not documented

[mcp292]

Please let me know if there is a forum specifically for Altair questions and help and I will move this there.

I haven't been able to find in the documentation how facet resolves multiple entries for the facetted parameter.

In the MWE below, what does facet() do when it encounters 3 entries where P = HD, and T = 0, yet only plots one value at HD, 0 in the t_facet (T) plot.

Is this just a flawed plot?

MWE

import pandas as pd
import altair as alt


data = [{"P": "HD", "T": 0,  "C":100, "Max": 1},
        {"P": "HD", "T": 0,  "C":400, "Max": 3},
        {"P": "HD", "T": 0,  "C":800, "Max": 0},
        {"P": "HD", "T": 23, "C":100, "Max": 2},
        {"P": "HD", "T": 23, "C":400, "Max": 1},
        {"P": "HD", "T": 23, "C":800, "Max": 1},
        {"P": "HD", "T": 80, "C":100, "Max": 7},
        {"P": "HD", "T": 80, "C":400, "Max": 2},
        {"P": "HD", "T": 80, "C":800, "Max": 1},
        {"P": "LD", "T": 0,  "C":100, "Max": 1},
        {"P": "LD", "T": 0,  "C":400, "Max": 2},
        {"P": "LD", "T": 0,  "C":800, "Max": 7},
        {"P": "LD", "T": 23, "C":100, "Max": 7},
        {"P": "LD", "T": 23, "C":400, "Max": 1},
        {"P": "LD", "T": 23, "C":800, "Max": 0},
        {"P": "LD", "T": 80, "C":100, "Max": 2},
        {"P": "LD", "T": 80, "C":400, "Max": 0},
        {"P": "LD", "T": 80, "C":800, "Max": 1}]

df = pd.DataFrame(data)

chart = alt.Chart(df).mark_bar().encode(x="P:N", y="Max:Q", color="P:N")

t_facet = chart.facet("T:N")
c_facet = chart.facet("C:N")

     P   T    C  Max
0   HD   0  100    1
1   HD   0  400    3
2   HD   0  800    0
3   HD  23  100    2
4   HD  23  400    1
5   HD  23  800    1
6   HD  80  100    7
7   HD  80  400    2
8   HD  80  800    1
9   LD   0  100    1
10  LD   0  400    2
11  LD   0  800    7
12  LD  23  100    7
13  LD  23  400    1
14  LD  23  800    0
15  LD  80  100    2
16  LD  80  400    0
17  LD  80  800    1

[jakevdp]

Because you haven't done any aggregation, in the location of duplicates the chart draws multiple bars plotted on top of each other, and the largest one dominates. You can see this better if you use mark_point() in place of mark_bar().

[mcp292]

Ah! Thank you! Where should I ask questions in the future?

My next question is how would you recommend displaying data of this dimension then?