# Plot with two DataFrames **URL:** \ **Category:** Questions\ **Tags:** nx, vegalite, tucan\ **Created:** [November 4, 2023, 11:02pm UTC](https://forum.elixirforum.com/t/plot-with-two-dataframes/59490 "2023-11-04T23:02:26Z")\ **Posts on this page:** 5\ **Page:** 1
**Author:** ![sehHeiden](https://forum.elixirforum.com/user_avatar/forum.elixirforum.com/sehheiden/32/31707_2.png) [@sehHeiden](https://forum.elixirforum.com/u/sehHeiden)\ **Post date:** [November 4, 2023, 11:02pm UTC](https://forum.elixirforum.com/t/plot-with-two-dataframes/59490/1 "2023-11-04T23:02:26Z")
I am searching for version to plot two dataframes with the same colon name into a layer each. I tried it in VegLite Json: ```json { "$schema": "https://vega.github.io/schema/vega-lite/v5.json", "data": { "values": [ {"x": 1, "y": 10}, {"x": 2, "y": 20}, {"x": 3, "y": 15} ], "name": "data1" }, "layer": [ { "data": { "values": [ {"x": 2, "y": 30}, {"x": 3, "y": 25}, {"x": 4, "y": 35} ], "name": "data2" }, "mark": "line", "encoding": { "x": {"field": "x", "type": "quantitative"}, "y": {"field": "y", "type": "quantitative"}, "color": {"value": "blue"} } }, { "data": {"name": "data1"}, "mark": "line", "transform": [ {"filter": {"field": "x", "range": [2, 3]}} ], "encoding": { "x": {"field": "x", "type": "quantitative"}, "y": {"field": "y", "type": "quantitative"}, "color": {"value": "red"} } } ], "resolve": { "scale": { "x": "shared", "y": "independent" } } } ``` What I came up with in Elixir draws the seconds DataFrame so that, it is WWWWAY to high over the correct place and the y range is evenly wrongly scaled: ```elixir def create_compare_graph(data_a, data_b, title, col) do Vl.new(width: 500, height: 300, title: title) |> Vl.data_from_values(data_a, only: ["day", col]) |> Vl.layers([ Vl.new() |> Vl.mark(:line, color: "blue") |> Vl.encode_field(:x, "day", type: :quantitative, title: "Day of the year" ) |> Vl.encode_field(:y, col, type: :quantitative, scale: [domain: [0, 1]], title: "DF1"), Vl.new() |>Vl.data_from_values(data_b, only: ["day", col]) |> Vl.mark(:line, color: "red") |> Vl.encode_field(:x, "day", type: :quantitative, title: "Day of the year") |> Vl.encode_field(:y, party, type: :quantitative, scale: [domain: [0, 1]], title: "DF2") ]) |> Vl.resolve(:scale, y: :independent) end end ``` Example Data: DF1: day value | 250 | 0.125 | 251 | 0.13 | 252 | 0.13 | 256 | 0.14 | 257 | 0.14 DF2: day value | 250 | 0.36 | 251 | 0.36 | 252 | 0.36 | 256 | 0.36 | 257 | 0.36 If possible I would prefer to do it, in Tucan as the code locks much cleaner with it. 😉 ---
**Author:** ![sehHeiden](https://forum.elixirforum.com/user_avatar/forum.elixirforum.com/sehheiden/32/31707_2.png) [@sehHeiden](https://forum.elixirforum.com/u/sehHeiden)\ **Post date:** [November 4, 2023, 11:24pm UTC](https://forum.elixirforum.com/t/plot-with-two-dataframes/59490/2 "2023-11-04T23:24:54Z")
A reply to myself. [VegaLite-datasets\_from\_values](https://hexdocs.pm/vega_lite/VegaLite.html#data_from_values/3) allows several sources. I tried to apply it with: ```elixir def create_compare_graph(data_a, data_b, title, col) do Vl.new(width: 500, height: 300, title: title) |> Vl.data_from_values(a: data_a, b: data_b, only: ["day", party]) |> Vl.layers([ Vl.new() |> Vl.data(name: "a") |> Vl.mark(:line, color: "blue") |> Vl.encode_field(:x, "day", type: :quantitative, title: "Day of the year" ) |> Vl.encode_field(:y, party, type: :quantitative, scale: [domain: [0, 1]], title: "Sentiment"), Vl.new() |> Vl.data(name: "b") |> Vl.mark(:line, color: "red") |> Vl.encode_field(:x, "day", type: :quantitative, title: "Day of the year") |> Vl.encode_field(:y, party, type: :quantitative, scale: [domain: [0, 1]], title: "Polls") ]) |> Vl.resolve(:scale, y: :independent) end end ``` Error: ```bash expected valid tabular data, but got ... (*Desribes the dataframe*) ``` Edited: the error Because chcnage Vl.Data → Vl.data ---
**Author:** ![awerment](https://forum.elixirforum.com/user_avatar/forum.elixirforum.com/awerment/32/30222_2.png) [@awerment](https://forum.elixirforum.com/u/awerment)\ **Post date:** [November 5, 2023, 7:56am UTC](https://forum.elixirforum.com/t/plot-with-two-dataframes/59490/3 "2023-11-05T07:56:04Z")
Are the domains in the real data significantly different? If so, does setting the scale resolution to `:shared` help? (`|> Vl.resolve(:scale, y: :shared`) Without a legend, independent scales can look confusing, although the chart is correct: ![independent_y_scale](https://forum.elixirforum.com/uploads/default/original/3X/6/b/6b3cb9f1387a77ce11a8225349ecab8fb97988c1.jpeg) With `:shared`: ![shared_y_scale](https://forum.elixirforum.com/uploads/default/original/3X/3/e/3ed2952fd7fb104a6361fb9f23589ff8c40e75cb.jpeg) ```elixir data_a = DF.new(day: 251..260, value: 1..10) data_b = DF.new(day: 251..260, value: 4..13) ``` ---
**Author:** ![sehHeiden](https://forum.elixirforum.com/user_avatar/forum.elixirforum.com/sehheiden/32/31707_2.png) [@sehHeiden](https://forum.elixirforum.com/u/sehHeiden)\ **Post date:** [November 5, 2023, 11:30am UTC](https://forum.elixirforum.com/t/plot-with-two-dataframes/59490/4 "2023-11-05T11:30:34Z")
Sorry, I did have a data error. And did not change the varible. Therefore they ware out of scale. I also applied Tucan. Which greatly reduces the amount of code. ```elixir def create_compare_graph(sentiment_data_source, poll_data_source, title, party) do Tucan.layers([ Tucan.lineplot(poll_data_source, "day", party, line_color: "red"), Tucan.lineplot(sentiment_data_source, "day", party, line_color: "green") ]) |> Tucan.set_width(500) |> Tucan.set_title(title) |> Tucan.Axes.set_x_title("Day of the year") |> Tucan.Axes.set_y_title("Sentiments, Polls") |> Tucan.Scale.set_y_domain(0,1) end ``` The only thing I would like to add is a Legend. ---
**Author:** ![sehHeiden](https://forum.elixirforum.com/user_avatar/forum.elixirforum.com/sehheiden/32/31707_2.png) [@sehHeiden](https://forum.elixirforum.com/u/sehHeiden)\ **Post date:** [November 5, 2023, 3:59pm UTC](https://forum.elixirforum.com/t/plot-with-two-dataframes/59490/5 "2023-11-05T15:59:10Z")
Solved it somewhat with Annotate.