Netanel Indik 311300784 Tal Komemi 311148902
Python 3.8
Rule # 0 info: empirical_error_on_test - 21.8 true_error_on_training - 7.75 Rule # 1 info: empirical_error_on_test - 22.55 true_error_on_training - 8.06 Rule # 2 info: empirical_error_on_test - 23.76 true_error_on_training - 9.08 Rule # 3 info: empirical_error_on_test - 24.45 true_error_on_training - 10.3 Rule # 4 info: empirical_error_on_test - 24.77 true_error_on_training - 10.29 Rule # 5 info: empirical_error_on_test - 24.96 true_error_on_training - 10.58 Rule # 6 info: empirical_error_on_test - 24.95 true_error_on_training - 10.34 Rule # 7 info: empirical_error_on_test - 24.98 true_error_on_training - 10.54
Rule # 0 info: empirical_error_on_test - 30.82 true_error_on_training - 9.93 Rule # 1 info: empirical_error_on_test - 30.92 true_error_on_training - 10.77 Rule # 2 info: empirical_error_on_test - 32.0 true_error_on_training - 10.78 Rule # 3 info: empirical_error_on_test - 31.63 true_error_on_training - 12.29 Rule # 4 info: empirical_error_on_test - 32.11 true_error_on_training - 11.3 Rule # 5 info: empirical_error_on_test - 31.37 true_error_on_training - 12.28 Rule # 6 info: empirical_error_on_test - 31.77 true_error_on_training - 11.21 Rule # 7 info: empirical_error_on_test - 31.23 true_error_on_training - 11.69
we've expected to see overfitting due to the fact that the VC dimension increases with the number of rules we combine, but for some reason it doesnt show in our results. more over, Iris data set showed better results with this model
We worked with the plugin "Code With Me" in PyCharm, while communicating with zoom We knew the code while working, so we did not rely on a particular source