散点图上的散景图回归线

时间:2019-02-09 06:36:35

标签: python regression bokeh scatter-plot

我使用Python和Bokeh在同一张图表中生成了两个散点图,并添加了复选框以允许单独查看散点图。

如何使用Bokeh为两个散点图(带有方程式)添加回归线?

output_file("Scatterplot.html")

#scatter plot
S0 = f.circle(A_area, A_price,
         fill_alpha=0.3, size=3, color='green')
S1 = f.circle(B_area, B_price,
         fill_alpha=0.3, size=3, color='blue')

#widget-checkbox
checkboxes = CheckboxGroup(labels=["A", "B"], active=[0, 1])
callback = CustomJS(code="""S0.visible = false; // same S0 passed in from args
                            S1.visible = false;
                            // cb_obj injected in by the callback
                            if (cb_obj.active.includes(0)){S0.visible = true;} // 0 index box is S0
                            if (cb_obj.active.includes(1)){S1.visible = true;}""",
                    args={'S0': S0, 'S1': S1})

checkboxes.js_on_click(callback)

2 个答案:

答案 0 :(得分:2)

Bokeh在其documentation中也有一个Slope类,用于绘制这样的回归线。您所需要做的就是截距和倾斜度。另外,我使用的是scikit-learn中的LinearRegression(),但Joris回答中的np.polyfit显然也可以工作。

import numpy as np
from sklearn.linear_model import LinearRegression
from bokeh.models import Slope
from bokeh.plotting import figure, show
from bokeh.io import output_notebook
output_notebook()

# Data
x=np.array([0,1,2,3,4,5,6,7,8])
y=np.array([1,2,3,5,4,6,8,7,9])

# Make and fit a linear regression model
model = LinearRegression().fit(x.reshape(-1, 1), y)
# x values need to be in a two-dimensional array, so use .reshape(-1, 1)

# Find the slope and intercept from the model
slope = model.coef_[0] # Takes the first element of the array
intercept = model.intercept_

# Make the regression line
regression_line = Slope(gradient=slope, y_intercept=intercept, line_color="red")

# Plot the data and regression line
fig=figure()
fig.circle(x, y)
fig.add_layout(regression_line)
show(fig)

答案 1 :(得分:0)

您先用numpy计算线条拟合,然后在bokeh中进行绘制:

import numpy as np
from bokeh.plotting import figure
from bokeh.io import show

#the data
x=np.array([0,1,2,3,4,5,6,7,8])
y=np.array([1,2,3,5,4,6,8,7,9])

# determine best fit line
par = np.polyfit(x, y, 1, full=True)
slope=par[0][0]
intercept=par[0][1]
y_predicted = [slope*i + intercept  for i in x]

# plot it
fig=figure()
fig.circle(x,y)
fig.line(x,y_predicted,color='red',legend='y='+str(round(slope,2))+'x+'+str(round(intercept,2)))
show(fig)

enter image description here