情节注释彼此太近（不可读）

Question

我有以下代码为 PCA 后的载荷创建图：

# Creating pipeline objects 
## PCA
pca = PCA(n_components=2)
## Create columntransformer to only scale a selected set of featues
categorical_ix = X.select_dtypes(exclude=np.number).columns

features = X.columns

ct = ColumnTransformer([
        ('encoder', OneHotEncoder(), categorical_ix),
        ('scaler', StandardScaler(), ['tenure', 'MonthlyCharges', 'TotalCharges'])
    ], remainder='passthrough')

# Create pipeline
pca_pipe = make_pipeline(ct,
                         pca)

# Fit data to pipeline
pca_result = pca_pipe.fit_transform(X)

loadings = pca.components_.T * np.sqrt(pca.explained_variance_)

fig = px.scatter(pca_result, x=0, y=1, color=customer_data_raw['Churn'])

for i, feature in enumerate(features):
    fig.add_shape(
        type='line',
        x0=0, y0=0,
        x1=loadings[i, 0],
        y1=loadings[i, 1]
    )
    fig.add_annotation(
        x=loadings[i, 0],
        y=loadings[i, 1],
        ax=0, ay=0,
        xanchor="center",
        yanchor="bottom",
        text=feature,
    )
fig.show()

产生以下输出：

如何使载荷标签可读？

编辑： X中有19个特征。

    gender  SeniorCitizen   Partner Dependents  tenure  PhoneService    MultipleLines   InternetService OnlineSecurity  OnlineBackup    DeviceProtection    TechSupport StreamingTV StreamingMovies Contract    PaperlessBilling    PaymentMethod   MonthlyCharges  TotalCharges
customerID                                                                          
7590-VHVEG  Female  0   Yes No  1   No  No phone service    DSL No  Yes No  No  No  No  Month-to-month  Yes Electronic check    29.85   29.85
5575-GNVDE  Male    0   No  No  34  Yes No  DSL Yes No  Yes No  No  No  One year    No  Mailed check    56.95   1889.50
3668-QPYBK  Male    0   No  No  2   Yes No  DSL Yes Yes No  No  No  No  Month-to-month  Yes Mailed check    53.85   108.15
7795-CFOCW  Male    0   No  No  45  No  No phone service    DSL Yes No  Yes Yes No  No  One year    No  Bank transfer (automatic)   42.30   1840.75
9237-HQITU  Female  0   No  No  2   Yes No  Fiber optic No  No  No  No  No  No  Month-to-month  Yes Electronic check    70.70   151.65

Answer 1

根据你的 DataFrame，你有 19 个特征，你将它们全部添加到你的行的位置，因为 ax 和 ay 都设置为 0。

我们可以更改 ax 和 ay，因为您循环遍历要旋转的特征，这有望使您的注释更容易区分。这是基于使用 x = r*cos(theta) 和 y = r*sin(theta) 从极坐标转换为笛卡尔坐标，其中 theta 通过值 0*360/19, 1*360/19, ... , 18*360/19。我们希望将 x 和 y 参考设置为 x 和 y 坐标而不是纸张坐标，然后设置 r=2 或与您的绘图相当的某个值（这将使注释线长度最长为 2）

from math import sin, cos, pi
r = 2 # this can be modified as needed, and is in units of the axis
theta = 2*pi/len(features)

for i, feature in enumerate(features):
    fig.add_shape(
        type='line',
        x0=0, y0=0,
        x1=loadings[i, 0],
        y1=loadings[i, 1]
    )
    fig.add_annotation(
        x=loadings[i, 0],
        y=loadings[i, 1],
        ax=r*sin(i*theta), 
        ay=r*cos(i*theta),
        axref="x",
        ayref="y",
        xanchor="center",
        yanchor="bottom",
        text=feature,
    )

情节注释彼此太近（不可读）

Plotly annotations too close to each other (not readable)

python

plotly

data-visualization

pca