Generalized Additive Models GAM in the StatsModels library
In this recipe, we explain what are generalized additive models in the StatsModels library
Recipe Objective - Generalized Additive Models (GAM) in the StatsModels library?
The generalized additive model allows a penalty estimation of the smooth terms of the generalized linear model.
For more related projects -
https://www.dezyre.com/projects/data-science-projects/deep-learning-projects
https://www.dezyre.com/projects/data-science-projects/neural-network-projects
Example:
# Importing libraries
import statsmodels.api as sm
from statsmodels.gam.api import GLMGam, BSplines
# Importing heart dataset from statsmodels in the form of pandas dataframe
data = sm.datasets.heart.load_pandas()
# Storing data
X = data.data
# Using BSplines for smoother
smoother = BSplines(X[['age','censors']], df=[10, 5], degree=[3, 3])
# Fit and summarize GLMGam model
model = GLMGam.from_formula("survival ~ age + censors", data=X, smoother=smoother)
model = model.fit()
# Model summary
model.summary()
Output -
Generalized Linear Model Regression Results
Dep. Variable: survival No. Observations: 69
Model: GLMGam Df Residuals: 58.00
Model Family: Gaussian Df Model: 10.00
Link Function: identity Scale: 1.7473e+05
Method: PIRLS Log-Likelihood: -508.37
Date: Mon, 22 Nov 2021 Deviance: 1.0134e+07
Time: 17:48:25 Pearson chi2: 1.01e+07
No. Iterations: 3
Covariance Type: nonrobust
coef std err z P>|z| [0.025 0.975]
Intercept 550.7731 626.940 0.879 0.380 -678.006 1779.552
age 3.1801 11.221 0.283 0.777 -18.812 25.172
censors -455.7141 120.386 -3.785 0.000 -691.666 -219.763
age_s0 -877.3193 795.380 -1.103 0.270 -2436.235 681.597
age_s1 468.7773 590.502 0.794 0.427 -688.586 1626.141
age_s2 59.9559 348.030 0.172 0.863 -622.170 742.082
age_s3 -31.0203 280.428 -0.111 0.912 -580.649 518.609
age_s4 124.3547 243.360 0.511 0.609 -352.623 601.332
age_s5 -203.0540 273.073 -0.744 0.457 -738.267 332.159
age_s6 -70.7493 564.662 -0.125 0.900 -1177.466 1035.967
age_s7 127.8988 747.057 0.171 0.864 -1336.307 1592.104
age_s8 -281.4048 389.188 -0.723 0.470 -1044.200 481.390
censors_s0 0 0 nan nan 0 0
censors_s1 0 0 nan nan 0 0
censors_s2 0 0 nan nan 0 0
censors_s3 0 0 nan nan 0 0
In this way, we can use Generalized Additive Models (GAM) in the StatsModels library.
What Users are saying..
Ray han
I think that they are fantastic. I attended Yale and Stanford and have worked at Honeywell,Oracle, and Arthur Andersen(Accenture) in the US. I have taken Big Data and Hadoop,NoSQL, Spark, Hadoop... Read More