LunchBox Linear Regression and Multivariate Regression vs. python results

Grasshopper
1

Hi all,

  1. While fitting a curve through linear regression method, using Grasshopper LunchBox, I noticed numbers do not match up with those calculated by python. However small, the discrepancy is really annoying.

test 02.gh (36.5 KB)
test 02.csv (1.6 KB)

01
01704×389 34.2 KB

  1. The second issue is the way input and output data are used in the template file provided by the plug-in developer. I suspect the input seeks “two or more” independent data—or predictors—whereas the output is where the dependent data is fed. The template shows otherwise; two output and one input.

02
02907×298 36.3 KB

The template, provided by the developer:
LBML_MultivariateLinearRegression.gh (25.0 KB)

Your suggestions and/or hints/solutions are highly appreciated.

2

Since the edit option is seemingly not available, I am gonna raise an alternative question in this regard.

This python code:

import pandas as pd
import statsmodels.api as sm
import seaborn as sns
import matplotlib.pyplot as plt

points = pd.read_csv("...3d points.csv")
model2 = sm.OLS.from_formula('y ~ x + z', data=points).fit()

print (model2.params)

sns.lmplot(x='x', y='y', palette='Blues', fit_reg=False, data=points)

plt.plot(points.x, model2.params[0]+model2.params[1]*points.x+model2.params[2]*-2, color='lightblue', linewidth=5)
plt.plot(points.x, model2.params[0]+model2.params[1]*points.x+model2.params[2]*4, color='blue', linewidth=5)
plt.plot(points.x, model2.params[0]+model2.params[1]*points.x+model2.params[2]*6, color='darkblue', linewidth=5)

plt.show()

Generates these results:

Intercept    18.172019
x             0.242383
z             1.625571

The grasshopper component of multivariate regression, however, provides different results (coefficients and the intercept).
LBML_MultivariateLinearRegression.gh (22.5 KB)
3d points.csv (2.1 KB)

How do we interpret this discrepancy?
if the question is fundamentally wrong, I would appreciate it if anyone could show me what the right question is like.