/////////////////////////////////////////// // Running SimpleGAN on imblearn_mammography /////////////////////////////////////////// Load 'data_input/imblearn_mammography' from imblearn non empty cut in data_input/imblearn_mammography! (7 points) Data loaded. -> Shuffling data ### Start exercise for synthetic point generator ====== Step 1/5 ======= -> Shuffling data -> Spliting data to slices ------ Step 1/5: Slice 1/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2095, 90 LR fn, tp: 13, 39 LR f1 score: 0.431 LR cohens kappa score: 0.411 LR average precision score: 0.608 -> test with 'GB' GB tn, fp: 2154, 31 GB fn, tp: 22, 30 GB f1 score: 0.531 GB cohens kappa score: 0.519 -> test with 'KNN' KNN tn, fp: 1483, 702 KNN fn, tp: 21, 31 KNN f1 score: 0.079 KNN cohens kappa score: 0.037 ------ Step 1/5: Slice 2/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2143, 42 LR fn, tp: 22, 30 LR f1 score: 0.484 LR cohens kappa score: 0.470 LR average precision score: 0.508 -> test with 'GB' GB tn, fp: 2184, 1 GB fn, tp: 24, 28 GB f1 score: 0.691 GB cohens kappa score: 0.686 -> test with 'KNN' KNN tn, fp: 2183, 2 KNN fn, tp: 30, 22 KNN f1 score: 0.579 KNN cohens kappa score: 0.573 ------ Step 1/5: Slice 3/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2068, 117 LR fn, tp: 11, 41 LR f1 score: 0.390 LR cohens kappa score: 0.368 LR average precision score: 0.536 -> test with 'GB' GB tn, fp: 2174, 11 GB fn, tp: 14, 38 GB f1 score: 0.752 GB cohens kappa score: 0.747 -> test with 'KNN' KNN tn, fp: 2175, 10 KNN fn, tp: 20, 32 KNN f1 score: 0.681 KNN cohens kappa score: 0.674 ------ Step 1/5: Slice 4/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2128, 57 LR fn, tp: 25, 27 LR f1 score: 0.397 LR cohens kappa score: 0.379 LR average precision score: 0.343 -> test with 'GB' GB tn, fp: 2176, 9 GB fn, tp: 26, 26 GB f1 score: 0.598 GB cohens kappa score: 0.590 -> test with 'KNN' KNN tn, fp: 2176, 9 KNN fn, tp: 31, 21 KNN f1 score: 0.512 KNN cohens kappa score: 0.504 ------ Step 1/5: Slice 5/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8532 synthetic samples -> test with 'LR' LR tn, fp: 2152, 31 LR fn, tp: 26, 26 LR f1 score: 0.477 LR cohens kappa score: 0.464 LR average precision score: 0.498 -> test with 'GB' GB tn, fp: 2174, 9 GB fn, tp: 23, 29 GB f1 score: 0.644 GB cohens kappa score: 0.637 -> test with 'KNN' KNN tn, fp: 1568, 615 KNN fn, tp: 27, 25 KNN f1 score: 0.072 KNN cohens kappa score: 0.031 ====== Step 2/5 ======= -> Shuffling data -> Spliting data to slices ------ Step 2/5: Slice 1/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2152, 33 LR fn, tp: 27, 25 LR f1 score: 0.455 LR cohens kappa score: 0.441 LR average precision score: 0.462 -> test with 'GB' GB tn, fp: 2178, 7 GB fn, tp: 25, 27 GB f1 score: 0.628 GB cohens kappa score: 0.621 -> test with 'KNN' KNN tn, fp: 2179, 6 KNN fn, tp: 22, 30 KNN f1 score: 0.682 KNN cohens kappa score: 0.676 ------ Step 2/5: Slice 2/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2071, 114 LR fn, tp: 17, 35 LR f1 score: 0.348 LR cohens kappa score: 0.325 LR average precision score: 0.453 -> test with 'GB' GB tn, fp: 2170, 15 GB fn, tp: 23, 29 GB f1 score: 0.604 GB cohens kappa score: 0.596 -> test with 'KNN' KNN tn, fp: 2180, 5 KNN fn, tp: 26, 26 KNN f1 score: 0.627 KNN cohens kappa score: 0.620 ------ Step 2/5: Slice 3/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2164, 21 LR fn, tp: 24, 28 LR f1 score: 0.554 LR cohens kappa score: 0.544 LR average precision score: 0.571 -> test with 'GB' GB tn, fp: 2179, 6 GB fn, tp: 26, 26 GB f1 score: 0.619 GB cohens kappa score: 0.612 -> test with 'KNN' KNN tn, fp: 1522, 663 KNN fn, tp: 28, 24 KNN f1 score: 0.065 KNN cohens kappa score: 0.023 ------ Step 2/5: Slice 4/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2163, 22 LR fn, tp: 22, 30 LR f1 score: 0.577 LR cohens kappa score: 0.567 LR average precision score: 0.606 -> test with 'GB' GB tn, fp: 2176, 9 GB fn, tp: 25, 27 GB f1 score: 0.614 GB cohens kappa score: 0.606 -> test with 'KNN' KNN tn, fp: 2180, 5 KNN fn, tp: 30, 22 KNN f1 score: 0.557 KNN cohens kappa score: 0.550 ------ Step 2/5: Slice 5/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8532 synthetic samples -> test with 'LR' LR tn, fp: 2132, 51 LR fn, tp: 20, 32 LR f1 score: 0.474 LR cohens kappa score: 0.459 LR average precision score: 0.478 -> test with 'GB' GB tn, fp: 2175, 8 GB fn, tp: 29, 23 GB f1 score: 0.554 GB cohens kappa score: 0.546 -> test with 'KNN' KNN tn, fp: 2177, 6 KNN fn, tp: 32, 20 KNN f1 score: 0.513 KNN cohens kappa score: 0.505 ====== Step 3/5 ======= -> Shuffling data -> Spliting data to slices ------ Step 3/5: Slice 1/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2119, 66 LR fn, tp: 13, 39 LR f1 score: 0.497 LR cohens kappa score: 0.481 LR average precision score: 0.657 -> test with 'GB' GB tn, fp: 2184, 1 GB fn, tp: 17, 35 GB f1 score: 0.795 GB cohens kappa score: 0.791 -> test with 'KNN' KNN tn, fp: 2182, 3 KNN fn, tp: 20, 32 KNN f1 score: 0.736 KNN cohens kappa score: 0.731 ------ Step 3/5: Slice 2/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2154, 31 LR fn, tp: 20, 32 LR f1 score: 0.557 LR cohens kappa score: 0.545 LR average precision score: 0.548 -> test with 'GB' GB tn, fp: 2173, 12 GB fn, tp: 27, 25 GB f1 score: 0.562 GB cohens kappa score: 0.553 -> test with 'KNN' KNN tn, fp: 2177, 8 KNN fn, tp: 30, 22 KNN f1 score: 0.537 KNN cohens kappa score: 0.529 ------ Step 3/5: Slice 3/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2162, 23 LR fn, tp: 24, 28 LR f1 score: 0.544 LR cohens kappa score: 0.533 LR average precision score: 0.603 -> test with 'GB' GB tn, fp: 2180, 5 GB fn, tp: 23, 29 GB f1 score: 0.674 GB cohens kappa score: 0.668 -> test with 'KNN' KNN tn, fp: 2181, 4 KNN fn, tp: 28, 24 KNN f1 score: 0.600 KNN cohens kappa score: 0.593 ------ Step 3/5: Slice 4/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2143, 42 LR fn, tp: 17, 35 LR f1 score: 0.543 LR cohens kappa score: 0.530 LR average precision score: 0.506 -> test with 'GB' GB tn, fp: 2177, 8 GB fn, tp: 25, 27 GB f1 score: 0.621 GB cohens kappa score: 0.613 -> test with 'KNN' KNN tn, fp: 2176, 9 KNN fn, tp: 26, 26 KNN f1 score: 0.598 KNN cohens kappa score: 0.590 ------ Step 3/5: Slice 5/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8532 synthetic samples -> test with 'LR' LR tn, fp: 2000, 183 LR fn, tp: 23, 29 LR f1 score: 0.220 LR cohens kappa score: 0.189 LR average precision score: 0.145 -> test with 'GB' GB tn, fp: 2174, 9 GB fn, tp: 28, 24 GB f1 score: 0.565 GB cohens kappa score: 0.557 -> test with 'KNN' KNN tn, fp: 2173, 10 KNN fn, tp: 30, 22 KNN f1 score: 0.524 KNN cohens kappa score: 0.515 ====== Step 4/5 ======= -> Shuffling data -> Spliting data to slices ------ Step 4/5: Slice 1/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2173, 12 LR fn, tp: 28, 24 LR f1 score: 0.545 LR cohens kappa score: 0.537 LR average precision score: 0.580 -> test with 'GB' GB tn, fp: 2179, 6 GB fn, tp: 25, 27 GB f1 score: 0.635 GB cohens kappa score: 0.629 -> test with 'KNN' KNN tn, fp: 2181, 4 KNN fn, tp: 27, 25 KNN f1 score: 0.617 KNN cohens kappa score: 0.611 ------ Step 4/5: Slice 2/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2174, 11 LR fn, tp: 23, 29 LR f1 score: 0.630 LR cohens kappa score: 0.623 LR average precision score: 0.622 -> test with 'GB' GB tn, fp: 2179, 6 GB fn, tp: 20, 32 GB f1 score: 0.711 GB cohens kappa score: 0.705 -> test with 'KNN' KNN tn, fp: 2181, 4 KNN fn, tp: 26, 26 KNN f1 score: 0.634 KNN cohens kappa score: 0.628 ------ Step 4/5: Slice 3/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2047, 138 LR fn, tp: 10, 42 LR f1 score: 0.362 LR cohens kappa score: 0.338 LR average precision score: 0.621 -> test with 'GB' GB tn, fp: 2168, 17 GB fn, tp: 16, 36 GB f1 score: 0.686 GB cohens kappa score: 0.678 -> test with 'KNN' KNN tn, fp: 2162, 23 KNN fn, tp: 22, 30 KNN f1 score: 0.571 KNN cohens kappa score: 0.561 ------ Step 4/5: Slice 4/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2154, 31 LR fn, tp: 25, 27 LR f1 score: 0.491 LR cohens kappa score: 0.478 LR average precision score: 0.496 -> test with 'GB' GB tn, fp: 2179, 6 GB fn, tp: 23, 29 GB f1 score: 0.667 GB cohens kappa score: 0.660 -> test with 'KNN' KNN tn, fp: 2180, 5 KNN fn, tp: 29, 23 KNN f1 score: 0.575 KNN cohens kappa score: 0.568 ------ Step 4/5: Slice 5/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8532 synthetic samples -> test with 'LR' LR tn, fp: 2160, 23 LR fn, tp: 26, 26 LR f1 score: 0.515 LR cohens kappa score: 0.504 LR average precision score: 0.490 -> test with 'GB' GB tn, fp: 2175, 8 GB fn, tp: 21, 31 GB f1 score: 0.681 GB cohens kappa score: 0.675 -> test with 'KNN' KNN tn, fp: 2177, 6 KNN fn, tp: 25, 27 KNN f1 score: 0.635 KNN cohens kappa score: 0.629 ====== Step 5/5 ======= -> Shuffling data -> Spliting data to slices ------ Step 5/5: Slice 1/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2158, 27 LR fn, tp: 17, 35 LR f1 score: 0.614 LR cohens kappa score: 0.604 LR average precision score: 0.629 -> test with 'GB' GB tn, fp: 2180, 5 GB fn, tp: 22, 30 GB f1 score: 0.690 GB cohens kappa score: 0.684 -> test with 'KNN' KNN tn, fp: 2180, 5 KNN fn, tp: 25, 27 KNN f1 score: 0.643 KNN cohens kappa score: 0.636 ------ Step 5/5: Slice 2/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2166, 19 LR fn, tp: 23, 29 LR f1 score: 0.580 LR cohens kappa score: 0.570 LR average precision score: 0.557 -> test with 'GB' GB tn, fp: 2176, 9 GB fn, tp: 25, 27 GB f1 score: 0.614 GB cohens kappa score: 0.606 -> test with 'KNN' KNN tn, fp: 1509, 676 KNN fn, tp: 27, 25 KNN f1 score: 0.066 KNN cohens kappa score: 0.024 ------ Step 5/5: Slice 3/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2112, 73 LR fn, tp: 15, 37 LR f1 score: 0.457 LR cohens kappa score: 0.439 LR average precision score: 0.549 -> test with 'GB' GB tn, fp: 2169, 16 GB fn, tp: 25, 27 GB f1 score: 0.568 GB cohens kappa score: 0.559 -> test with 'KNN' KNN tn, fp: 2169, 16 KNN fn, tp: 26, 26 KNN f1 score: 0.553 KNN cohens kappa score: 0.544 ------ Step 5/5: Slice 4/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8530 synthetic samples -> test with 'LR' LR tn, fp: 2114, 71 LR fn, tp: 12, 40 LR f1 score: 0.491 LR cohens kappa score: 0.474 LR average precision score: 0.592 -> test with 'GB' GB tn, fp: 2164, 21 GB fn, tp: 15, 37 GB f1 score: 0.673 GB cohens kappa score: 0.665 -> test with 'KNN' KNN tn, fp: 2173, 12 KNN fn, tp: 18, 34 KNN f1 score: 0.694 KNN cohens kappa score: 0.687 ------ Step 5/5: Slice 5/5 ------- -> Reset the GAN -> Train generator for synthetic samples Epoch 1/3 Epoch 2/3 Epoch 3/3 -> create 8532 synthetic samples -> test with 'LR' LR tn, fp: 2169, 14 LR fn, tp: 24, 28 LR f1 score: 0.596 LR cohens kappa score: 0.587 LR average precision score: 0.614 -> test with 'GB' GB tn, fp: 2175, 8 GB fn, tp: 22, 30 GB f1 score: 0.667 GB cohens kappa score: 0.660 -> test with 'KNN' KNN tn, fp: 2180, 3 KNN fn, tp: 28, 24 KNN f1 score: 0.608 KNN cohens kappa score: 0.601 ### Exercise is done. -----[ LR ]----- maximum: LR tn, fp: 2174, 183 LR fn, tp: 28, 42 LR f1 score: 0.630 LR cohens kappa score: 0.623 LR average precision score: 0.657 average: LR tn, fp: 2130.92, 53.68 LR fn, tp: 20.28, 31.72 LR f1 score: 0.489 LR cohens kappa score: 0.474 LR average precision score: 0.531 minimum: LR tn, fp: 2000, 11 LR fn, tp: 10, 24 LR f1 score: 0.220 LR cohens kappa score: 0.189 LR average precision score: 0.145 -----[ GB ]----- maximum: GB tn, fp: 2184, 31 GB fn, tp: 29, 38 GB f1 score: 0.795 GB cohens kappa score: 0.791 average: GB tn, fp: 2174.88, 9.72 GB fn, tp: 22.84, 29.16 GB f1 score: 0.642 GB cohens kappa score: 0.635 minimum: GB tn, fp: 2154, 1 GB fn, tp: 14, 23 GB f1 score: 0.531 GB cohens kappa score: 0.519 -----[ KNN ]----- maximum: KNN tn, fp: 2183, 702 KNN fn, tp: 32, 34 KNN f1 score: 0.736 KNN cohens kappa score: 0.731 average: KNN tn, fp: 2072.16, 112.44 KNN fn, tp: 26.16, 25.84 KNN f1 score: 0.518 KNN cohens kappa score: 0.506 minimum: KNN tn, fp: 1483, 2 KNN fn, tp: 18, 20 KNN f1 score: 0.065 KNN cohens kappa score: 0.023