I have a small dataset (1500 rows) and to predict the imbalanced target, I am running two linear models (linear regression and lasso) and one nonlinear model (Neural Network) on it. I am using Area Under Precision Recall Curve (AUPRC) to compare the three models. The baseline in the curve is 10%, AUC for linear regression is 11%, AUC for lasso is 11.2%, and AUC for NN is 11.35%. Can I say that the learning models have improved the random guessing? Is the difference between lasso and NN (11.2% and 11.35%) enough to say the relationships are nonlinear rather than linear? Is there any way (test) to show that the scores are significantly different (such as t-test or p-values)?
Asked
Active
Viewed 104 times
1
-
Is this a Classification or Regression problem? I believe you meant LogisticRegression? – 10xAI Mar 05 '21 at 13:32
-
Yes, logistic regression. – Shahab Kazemi Mar 05 '21 at 14:42
-
please explain AUPRC ? what purpose does it serve ? Similar values for each of three approaches simply indicates that these approaches are overlapping - based on similar principles. – Subhash C. Davar Mar 13 '21 at 00:00
-
"The baseline in the curve is 10%" . Does it mean a threshold ? – Subhash C. Davar Mar 13 '21 at 00:10
-
please explain your data/scores - continuous or Nominal or ratio. – Subhash C. Davar Mar 13 '21 at 00:19