Fitted regression line parameters in python

Question

How can I get the intercept and the slope of the fitted regression line in qqplot? Here’s a small working example, where I want the parameters of the red regression line:

import statsmodels.api as sm
import scipy.stats as stats
import numpy as np

np.random.seed(100)
a = np.random.normal(0, 4, 100)
sm.qqplot(a, stats.norm(loc=5, scale=1), line="r")

Asked By: statwoman

||

Source

Answer 1

You can extract the x,y coordinates of the line and use numpy polyfit to get your slope and intercept.

import statsmodels.api as sm
import scipy.stats as stats
import numpy as np

np.random.seed(100)
a = np.random.normal(0, 4, 100)
fig = sm.qqplot(a, stats.norm(loc=5, scale=1), line="r")

line = fig.gca().lines[-1]

slope, intercept =  np.polyfit(line.get_xdata(),line.get_ydata(),1)

print(slope, intercept)

Output

4.027047644266026 -20.55190163670639

Answered By: Chris

Answer 2

Use the returned figure to grab the relevant axes and line and calculate slope and intercept from the xydata. For reference, fig.axes[0].lines[0] is the data, lines[1] is the fit.

import statsmodels.api as sm
import scipy.stats as stats
import numpy as np

np.random.seed(100)
a = np.random.normal(0, 4, 100)
fig = sm.qqplot(a, stats.norm(loc=5, scale=1), line="r")


# Grab line data from figure
xydata = fig.axes[0].lines[1].get_xydata()
diff = (xydata - np.roll(xydata, 1, axis = 0))[1:]
slopes = diff[:, 1]/diff[:, 0]
slope = slopes.mean()

y_ints = xydata[:, 1] - slope*xydata[:, 0]
y_int = y_ints.mean()

# Re-plot line for visual confirmation
plt.plot(xydata[:, 0], xydata[:, 0]*slope + y_int, 'k--')

Answered By: Michael Cao

Fitted regression line parameters in python

Question:

Answers: