Plotly python：多项选择交互式绘图

Posted 2023-02-18

【中文标题】Plotly python：多项选择交互式绘图

【英文标题】：Plotly python: Multiple item selection interactive plot

【发布时间】：2021-10-24 19:02:39

【问题描述】：

我希望能够选择多个项目，而不是仅选择一项或全部。到目前为止，我只能选择一项并显示其图表，或者选择“全部” 并显示所有图表。

我希望能够选择 2 个或更多项目并将它们的图显示在如下所示的单个图表中。

我是 plotly 的新手，因此非常感谢任何帮助。非常感谢:)

我的代码：

def interactive_multi_plot(actual, forecast_1, forecast_2, title, addAll = True):
fig = go.Figure()
    

for column in forecast_1.columns.to_list():
    fig.add_trace(
        go.Scatter(
            x = forecast_1.index,
            y = forecast_1[column],
            name = "Forecast_SI"
        )

    )

    
    button_all = dict(label = 'All',
                  method = 'update',
                  args = ['visible': forecast_1.columns.isin(forecast_1.columns),
                           'title': 'All',
                           'showlegend':True])
    
for column in forecast_2.columns.to_list():
    fig.add_trace(
        go.Scatter(
            x = forecast_2.index,
            y = forecast_2[column],
            name = "Forecast_LSTM" 
        )

    )

    
    button_all = dict(label = 'All',
                  method = 'update',
                  args = ['visible': forecast_2.columns.isin(forecast_2.columns),
                           'title': 'All',
                           'showlegend':True])
for column in actual.columns.to_list():
    fig.add_trace(
        go.Scatter(
            x = actual.index,
            y = actual[column],
            name = "True values" 
        )

    )

    
    button_all = dict(label = 'All',
                  method = 'update',
                  args = ['visible': actual.columns.isin(actual.columns),
                           'title': 'All',
                           'showlegend':True])
    
fig.layout.plot_bgcolor = '#010028'
fig.layout.paper_bgcolor = '#010028'
def create_layout_button(column):
    return dict(label = column,
                method = 'update',
                args = ['visible': actual.columns.isin([column]),
                         'title': column,
                         'showlegend': True])
fig.update_layout(
    updatemenus=[go.layout.Updatemenu(
        active = 0,
        buttons = ([button_all] * addAll) +  list(actual.columns.map(lambda column: create_layout_button(column)))
        )
    ]     
)
# Update remaining layout properties
fig.update_layout(
    title_text=title,
    height=800,
    font = dict(color='#fff', size=12)
)


fig.show()

【参考方案1】：

这在 dash 中很简单，因为 https://dash.plotly.com/dash-core-components/dropdown 支持 多选和搜索

设置

对您的功能进行了微小的更改。定义跟踪的 meta 属性以供以后在 dash 回调中使用 模拟一些未提供的数据。三个数据框（实际、预测 1 和预测 2） 使用修改后的代码创建图形

集成到

dash

应用

设置

import pandas as pd
import numpy as np
import plotly.graph_objects as go

def interactive_multi_plot(actual, forecast_1, forecast_2, title, addAll = True):
    fig = go.Figure()
    for column in forecast_1.columns.to_list():
        fig.add_trace(
            go.Scatter(x = forecast_1.index,y = forecast_1[column],name = "Forecast_SI",meta = column))

    for column in forecast_2.columns.to_list():
        fig.add_trace(
            go.Scatter(x = forecast_2.index,y = forecast_2[column],name = "Forecast_LSTM" ,meta = column))
    for column in actual.columns.to_list():
        fig.add_trace(
            go.Scatter(x = actual.index,y = actual[column],name = "True values" ,meta = column))

    fig.layout.plot_bgcolor = '#010028'
    fig.layout.paper_bgcolor = '#010028'

    #  Update remaining layout properties
    fig.update_layout(
        title_text=title,
        height=800,
        font = dict(color='#fff', size=12)
    )

    return fig

S = 100
C = 10

actual = pd.DataFrame(
        c: np.sort(np.random.uniform(0, 600, S))
        for c in [f"ab-c" for a, b, c in zip(np.random.randint(100, 200, C),
                                                   np.random.choice(list("ABCDEF"), C),
                                                   np.random.randint(300, 400, C),)]
)

f1 = actual.assign(**c:actual[c]*1.1 for c in actual.columns)
f2 = actual.assign(**c:actual[c]*1.2 for c in actual.columns)

fig = interactive_multi_plot(actual, f1, f2, "Orders")

dash 应用

import dash_core_components as dcc
import dash_html_components as html
from dash.dependencies import Input, Output, State
from jupyter_dash import JupyterDash

# Build App
app = JupyterDash(__name__)
app.layout = html.Div(
    [
        dcc.Dropdown(
            id="lines",
            options=["label": c, "value": c for c in actual.columns.tolist()],
            value=None,
            multi=True
        ),
        dcc.Graph(id="interactive-multiplot", figure=fig),
    ]
)

@app.callback(
    Output("interactive-multiplot", "figure"),
    Input("lines", "value"),
    State("interactive-multiplot", "figure"),
)
def updateGraphCB(lines, fig):
    # filter traces...
    fig = go.Figure(fig).update_traces(visible=False)
    for line in [] if not lines else lines:
        fig.update_traces(visible=True, selector="meta":line)
    return fig

app.run_server(mode="inline")

【讨论】：

你好罗！你能解释一下你在代码的设置部分做什么吗？提前谢谢！

您好！谢谢您的帮助！我不明白为什么现在这些线是线性的，而且图表看起来与原始图表完全不同。另外，为什么项目名称不同，为什么随机选择？谢谢！

随机的唯一用途是生成数据帧，与绘图无关。你没有提供样本数据，所以我模拟了它，所以如果你想让我调查，我可以拨打这个电话fig = interactive_multi_plot(actual, f1, f2, "Orders") 最简单的方法是提供所有三个数据帧的样本数据，所以我不必假设它可能或可能如何不是结构化的