无论如何平滑一个情节折线图？

Question

我有很多图，我试图将其绘制为平滑曲线而不是带有易变标记的线图，是否有任何选择可以做到这一点？下面是一个代表其中一个图的示例字典，然后我将该字典转换为 DataFrame 并将其放入 plotly 中。问题是，情节看起来不太干净（波动很大），我想要一些可以平滑它的东西，有没有办法做到这一点？

    uncert = {0.01: 0.3132940811899597,
     0.03: 0.3654332700822265,
     0.05: 0.26783984501130126,
     0.07: 0.4321293089229754,
     0.09: 0.25012641159615706,
     0.11: 0.3714236470182696,
     0.13: 0.38346262341325815,
     0.15: 0.35005413208228076,
     0.17: 0.5588615782664942,
     0.19: 0.5836015906078394,
     0.21: 0.5266019417475728,
     0.23: 0.6645418326693228,
     0.25: 0.6699386503067485,
     0.27: 0.6684177348182391,
     0.29: 0.711600777705768,
     0.31: 0.7152067585593596,
     0.33: 0.6994047619047619,
     0.35: 0.6908919301557338,
     0.37: 0.7428780131482835,
     0.39: 0.6894644204174001,
     0.41: 0.7527301092043682,
     0.43: 0.816200215285253,
     0.45: 0.8000557724484105,
     0.47: 0.7623733719247467,
     0.49: 0.843609022556391,
     0.5: 0.7963190184049078,
     0.52: 0.8063279002876317,
     0.54: 0.8296098699566522,
     0.56: 0.8319386331938632,
     0.58: 0.7823228634039445,
     0.6: 0.7898773006134969,
     0.62: 0.8312474767864352,
     0.64: 0.8414997869620793,
     0.66: 0.8583032490974728,
     0.68: 0.8475551294343241,
     0.7: 0.8271983640081799,
     0.72: 0.8509589041095891,
     0.74: 0.848377581120944} 


plot = pd.DataFrame({"uncert":uncert})
fig = px.line(plot, x='% data', y=plot.columns[1:], markers=True, title="")

Answer 1

您可以使用任何您想要的分析模型并将其插入到迹线的 y 值中
展示了两个：
1. https://docs.scipy.org/doc/scipy/reference/generated/scipy.signal.savgol_filter.html
2. https://www.statsmodels.org/devel/generated/statsmodels.nonparametric.smoothers_lowess.lowess.html

from scipy import signal
import pandas as pd
import plotly.express as px
import statsmodels.api as sm

uncert = {
    0.01: 0.3132940811899597,
    0.03: 0.3654332700822265,
    0.05: 0.26783984501130126,
    0.07: 0.4321293089229754,
    0.09: 0.25012641159615706,
    0.11: 0.3714236470182696,
    0.13: 0.38346262341325815,
    0.15: 0.35005413208228076,
    0.17: 0.5588615782664942,
    0.19: 0.5836015906078394,
    0.21: 0.5266019417475728,
    0.23: 0.6645418326693228,
    0.25: 0.6699386503067485,
    0.27: 0.6684177348182391,
    0.29: 0.711600777705768,
    0.31: 0.7152067585593596,
    0.33: 0.6994047619047619,
    0.35: 0.6908919301557338,
    0.37: 0.7428780131482835,
    0.39: 0.6894644204174001,
    0.41: 0.7527301092043682,
    0.43: 0.816200215285253,
    0.45: 0.8000557724484105,
    0.47: 0.7623733719247467,
    0.49: 0.843609022556391,
    0.5: 0.7963190184049078,
    0.52: 0.8063279002876317,
    0.54: 0.8296098699566522,
    0.56: 0.8319386331938632,
    0.58: 0.7823228634039445,
    0.6: 0.7898773006134969,
    0.62: 0.8312474767864352,
    0.64: 0.8414997869620793,
    0.66: 0.8583032490974728,
    0.68: 0.8475551294343241,
    0.7: 0.8271983640081799,
    0.72: 0.8509589041095891,
    0.74: 0.848377581120944,
}


plot = pd.DataFrame({"uncert": uncert})
fig = (
    px.line(plot, y="uncert", markers=True, title="")
    .update_traces(name="original")
    .add_traces(
        px.line(
            plot,
            y=signal.savgol_filter(
                plot["uncert"], 11, 3
            ),  # window size used for filtering
            markers=True,
        )
        .update_traces(name="savgol")
        .data
    )
    .add_traces(
        px.line(
            plot,
            y=sm.nonparametric.lowess(plot["uncert"], plot.index, frac=0.3)[
                :, 1
            ],  # window size used for filtering
            markers=True,
        )
        .update_traces(name="lowres")
        .data
    )
    .update_traces(showlegend=True, line_color=None)
)

fig

无论如何平滑一个情节折线图？

Anyway to smooth a plotly line chart?

python

plot

plotly