/////////////////////////////////////////// // Running convGAN-majority-5 on kaggle_creditcard /////////////////////////////////////////// Load 'data_input/kaggle_creditcard' 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 -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 56164, 699 GAN fn, tp: 36, 63 GAN f1 score: 0.146 GAN cohens kappa score: 0.144 -> test with 'LR' LR tn, fp: 53960, 2903 LR fn, tp: 16, 83 LR f1 score: 0.054 LR cohens kappa score: 0.051 LR average precision score: 0.568 -> test with 'GB' GB tn, fp: 56550, 313 GB fn, tp: 19, 80 GB f1 score: 0.325 GB cohens kappa score: 0.323 -> test with 'KNN' KNN tn, fp: 56654, 209 KNN fn, tp: 78, 21 KNN f1 score: 0.128 KNN cohens kappa score: 0.126 ------ Step 1/5: Slice 2/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 801, 56062 GAN fn, tp: 0, 99 GAN f1 score: 0.004 GAN cohens kappa score: 0.000 -> test with 'LR' LR tn, fp: 53519, 3344 LR fn, tp: 6, 93 LR f1 score: 0.053 LR cohens kappa score: 0.049 LR average precision score: 0.711 -> test with 'GB' GB tn, fp: 56592, 271 GB fn, tp: 10, 89 GB f1 score: 0.388 GB cohens kappa score: 0.386 -> test with 'KNN' KNN tn, fp: 56676, 187 KNN fn, tp: 81, 18 KNN f1 score: 0.118 KNN cohens kappa score: 0.116 ------ Step 1/5: Slice 3/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 39860, 17003 GAN fn, tp: 3, 96 GAN f1 score: 0.011 GAN cohens kappa score: 0.008 -> test with 'LR' LR tn, fp: 55010, 1853 LR fn, tp: 9, 90 LR f1 score: 0.088 LR cohens kappa score: 0.085 LR average precision score: 0.683 -> test with 'GB' GB tn, fp: 56562, 301 GB fn, tp: 13, 86 GB f1 score: 0.354 GB cohens kappa score: 0.352 -> test with 'KNN' KNN tn, fp: 56536, 327 KNN fn, tp: 76, 23 KNN f1 score: 0.102 KNN cohens kappa score: 0.100 ------ Step 1/5: Slice 4/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 56745, 118 GAN fn, tp: 43, 56 GAN f1 score: 0.410 GAN cohens kappa score: 0.409 -> test with 'LR' LR tn, fp: 55193, 1670 LR fn, tp: 6, 93 LR f1 score: 0.100 LR cohens kappa score: 0.097 LR average precision score: 0.752 -> test with 'GB' GB tn, fp: 56514, 349 GB fn, tp: 8, 91 GB f1 score: 0.338 GB cohens kappa score: 0.336 -> test with 'KNN' KNN tn, fp: 56516, 347 KNN fn, tp: 71, 28 KNN f1 score: 0.118 KNN cohens kappa score: 0.116 ------ Step 1/5: Slice 5/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227056 synthetic samples -> test with GAN.predict GAN tn, fp: 991, 55872 GAN fn, tp: 0, 96 GAN f1 score: 0.003 GAN cohens kappa score: 0.000 -> test with 'LR' LR tn, fp: 55266, 1597 LR fn, tp: 7, 89 LR f1 score: 0.100 LR cohens kappa score: 0.097 LR average precision score: 0.854 -> test with 'GB' GB tn, fp: 56509, 354 GB fn, tp: 10, 86 GB f1 score: 0.321 GB cohens kappa score: 0.319 -> test with 'KNN' KNN tn, fp: 56483, 380 KNN fn, tp: 71, 25 KNN f1 score: 0.100 KNN cohens kappa score: 0.097 ====== Step 2/5 ======= -> Shuffling data -> Spliting data to slices ------ Step 2/5: Slice 1/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 27129, 29734 GAN fn, tp: 1, 98 GAN f1 score: 0.007 GAN cohens kappa score: 0.003 -> test with 'LR' LR tn, fp: 54102, 2761 LR fn, tp: 8, 91 LR f1 score: 0.062 LR cohens kappa score: 0.059 LR average precision score: 0.719 -> test with 'GB' GB tn, fp: 56469, 394 GB fn, tp: 10, 89 GB f1 score: 0.306 GB cohens kappa score: 0.304 -> test with 'KNN' KNN tn, fp: 56476, 387 KNN fn, tp: 75, 24 KNN f1 score: 0.094 KNN cohens kappa score: 0.092 ------ Step 2/5: Slice 2/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 55309, 1554 GAN fn, tp: 13, 86 GAN f1 score: 0.099 GAN cohens kappa score: 0.096 -> test with 'LR' LR tn, fp: 54067, 2796 LR fn, tp: 12, 87 LR f1 score: 0.058 LR cohens kappa score: 0.055 LR average precision score: 0.611 -> test with 'GB' GB tn, fp: 56443, 420 GB fn, tp: 11, 88 GB f1 score: 0.290 GB cohens kappa score: 0.288 -> test with 'KNN' KNN tn, fp: 56541, 322 KNN fn, tp: 71, 28 KNN f1 score: 0.125 KNN cohens kappa score: 0.122 ------ Step 2/5: Slice 3/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 56670, 193 GAN fn, tp: 49, 50 GAN f1 score: 0.292 GAN cohens kappa score: 0.291 -> test with 'LR' LR tn, fp: 54617, 2246 LR fn, tp: 10, 89 LR f1 score: 0.073 LR cohens kappa score: 0.070 LR average precision score: 0.653 -> test with 'GB' GB tn, fp: 56525, 338 GB fn, tp: 12, 87 GB f1 score: 0.332 GB cohens kappa score: 0.330 -> test with 'KNN' KNN tn, fp: 56435, 428 KNN fn, tp: 66, 33 KNN f1 score: 0.118 KNN cohens kappa score: 0.115 ------ Step 2/5: Slice 4/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 56631, 232 GAN fn, tp: 29, 70 GAN f1 score: 0.349 GAN cohens kappa score: 0.347 -> test with 'LR' LR tn, fp: 55314, 1549 LR fn, tp: 7, 92 LR f1 score: 0.106 LR cohens kappa score: 0.103 LR average precision score: 0.719 -> test with 'GB' GB tn, fp: 56564, 299 GB fn, tp: 12, 87 GB f1 score: 0.359 GB cohens kappa score: 0.357 -> test with 'KNN' KNN tn, fp: 56575, 288 KNN fn, tp: 78, 21 KNN f1 score: 0.103 KNN cohens kappa score: 0.101 ------ Step 2/5: Slice 5/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227056 synthetic samples -> test with GAN.predict GAN tn, fp: 56719, 144 GAN fn, tp: 40, 56 GAN f1 score: 0.378 GAN cohens kappa score: 0.377 -> test with 'LR' LR tn, fp: 54929, 1934 LR fn, tp: 11, 85 LR f1 score: 0.080 LR cohens kappa score: 0.077 LR average precision score: 0.753 -> test with 'GB' GB tn, fp: 56466, 397 GB fn, tp: 14, 82 GB f1 score: 0.285 GB cohens kappa score: 0.283 -> test with 'KNN' KNN tn, fp: 56696, 167 KNN fn, tp: 76, 20 KNN f1 score: 0.141 KNN cohens kappa score: 0.139 ====== Step 3/5 ======= -> Shuffling data -> Spliting data to slices ------ Step 3/5: Slice 1/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 55199, 1664 GAN fn, tp: 14, 85 GAN f1 score: 0.092 GAN cohens kappa score: 0.089 -> test with 'LR' LR tn, fp: 54138, 2725 LR fn, tp: 14, 85 LR f1 score: 0.058 LR cohens kappa score: 0.055 LR average precision score: 0.617 -> test with 'GB' GB tn, fp: 56598, 265 GB fn, tp: 15, 84 GB f1 score: 0.375 GB cohens kappa score: 0.373 -> test with 'KNN' KNN tn, fp: 56453, 410 KNN fn, tp: 73, 26 KNN f1 score: 0.097 KNN cohens kappa score: 0.095 ------ Step 3/5: Slice 2/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 56534, 329 GAN fn, tp: 42, 57 GAN f1 score: 0.235 GAN cohens kappa score: 0.233 -> test with 'LR' LR tn, fp: 54318, 2545 LR fn, tp: 10, 89 LR f1 score: 0.065 LR cohens kappa score: 0.062 LR average precision score: 0.600 -> test with 'GB' GB tn, fp: 56555, 308 GB fn, tp: 11, 88 GB f1 score: 0.356 GB cohens kappa score: 0.354 -> test with 'KNN' KNN tn, fp: 56453, 410 KNN fn, tp: 71, 28 KNN f1 score: 0.104 KNN cohens kappa score: 0.102 ------ Step 3/5: Slice 3/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 56114, 749 GAN fn, tp: 15, 84 GAN f1 score: 0.180 GAN cohens kappa score: 0.178 -> test with 'LR' LR tn, fp: 55387, 1476 LR fn, tp: 11, 88 LR f1 score: 0.106 LR cohens kappa score: 0.103 LR average precision score: 0.721 -> test with 'GB' GB tn, fp: 56653, 210 GB fn, tp: 14, 85 GB f1 score: 0.431 GB cohens kappa score: 0.430 -> test with 'KNN' KNN tn, fp: 56285, 578 KNN fn, tp: 71, 28 KNN f1 score: 0.079 KNN cohens kappa score: 0.077 ------ Step 3/5: Slice 4/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 50741, 6122 GAN fn, tp: 6, 93 GAN f1 score: 0.029 GAN cohens kappa score: 0.026 -> test with 'LR' LR tn, fp: 55113, 1750 LR fn, tp: 9, 90 LR f1 score: 0.093 LR cohens kappa score: 0.090 LR average precision score: 0.796 -> test with 'GB' GB tn, fp: 56531, 332 GB fn, tp: 9, 90 GB f1 score: 0.345 GB cohens kappa score: 0.344 -> test with 'KNN' KNN tn, fp: 56470, 393 KNN fn, tp: 76, 23 KNN f1 score: 0.089 KNN cohens kappa score: 0.087 ------ Step 3/5: Slice 5/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227056 synthetic samples -> test with GAN.predict GAN tn, fp: 56592, 271 GAN fn, tp: 28, 68 GAN f1 score: 0.313 GAN cohens kappa score: 0.311 -> test with 'LR' LR tn, fp: 55327, 1536 LR fn, tp: 10, 86 LR f1 score: 0.100 LR cohens kappa score: 0.097 LR average precision score: 0.760 -> test with 'GB' GB tn, fp: 56555, 308 GB fn, tp: 12, 84 GB f1 score: 0.344 GB cohens kappa score: 0.342 -> test with 'KNN' KNN tn, fp: 56420, 443 KNN fn, tp: 66, 30 KNN f1 score: 0.105 KNN cohens kappa score: 0.103 ====== Step 4/5 ======= -> Shuffling data -> Spliting data to slices ------ Step 4/5: Slice 1/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 23078, 33785 GAN fn, tp: 3, 96 GAN f1 score: 0.006 GAN cohens kappa score: 0.002 -> test with 'LR' LR tn, fp: 54806, 2057 LR fn, tp: 5, 94 LR f1 score: 0.084 LR cohens kappa score: 0.081 LR average precision score: 0.702 -> test with 'GB' GB tn, fp: 56518, 345 GB fn, tp: 8, 91 GB f1 score: 0.340 GB cohens kappa score: 0.338 -> test with 'KNN' KNN tn, fp: 56571, 292 KNN fn, tp: 77, 22 KNN f1 score: 0.107 KNN cohens kappa score: 0.104 ------ Step 4/5: Slice 2/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 53630, 3233 GAN fn, tp: 11, 88 GAN f1 score: 0.051 GAN cohens kappa score: 0.048 -> test with 'LR' LR tn, fp: 54935, 1928 LR fn, tp: 12, 87 LR f1 score: 0.082 LR cohens kappa score: 0.079 LR average precision score: 0.651 -> test with 'GB' GB tn, fp: 56553, 310 GB fn, tp: 12, 87 GB f1 score: 0.351 GB cohens kappa score: 0.349 -> test with 'KNN' KNN tn, fp: 56392, 471 KNN fn, tp: 80, 19 KNN f1 score: 0.065 KNN cohens kappa score: 0.062 ------ Step 4/5: Slice 3/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 56154, 709 GAN fn, tp: 19, 80 GAN f1 score: 0.180 GAN cohens kappa score: 0.178 -> test with 'LR' LR tn, fp: 54382, 2481 LR fn, tp: 15, 84 LR f1 score: 0.063 LR cohens kappa score: 0.060 LR average precision score: 0.684 -> test with 'GB' GB tn, fp: 56479, 384 GB fn, tp: 12, 87 GB f1 score: 0.305 GB cohens kappa score: 0.303 -> test with 'KNN' KNN tn, fp: 56490, 373 KNN fn, tp: 69, 30 KNN f1 score: 0.120 KNN cohens kappa score: 0.117 ------ Step 4/5: Slice 4/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 41568, 15295 GAN fn, tp: 2, 97 GAN f1 score: 0.013 GAN cohens kappa score: 0.009 -> test with 'LR' LR tn, fp: 54841, 2022 LR fn, tp: 7, 92 LR f1 score: 0.083 LR cohens kappa score: 0.080 LR average precision score: 0.762 -> test with 'GB' GB tn, fp: 56487, 376 GB fn, tp: 11, 88 GB f1 score: 0.313 GB cohens kappa score: 0.311 -> test with 'KNN' KNN tn, fp: 56549, 314 KNN fn, tp: 66, 33 KNN f1 score: 0.148 KNN cohens kappa score: 0.146 ------ Step 4/5: Slice 5/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227056 synthetic samples -> test with GAN.predict GAN tn, fp: 54292, 2571 GAN fn, tp: 10, 86 GAN f1 score: 0.062 GAN cohens kappa score: 0.059 -> test with 'LR' LR tn, fp: 54314, 2549 LR fn, tp: 7, 89 LR f1 score: 0.065 LR cohens kappa score: 0.062 LR average precision score: 0.737 -> test with 'GB' GB tn, fp: 56610, 253 GB fn, tp: 15, 81 GB f1 score: 0.377 GB cohens kappa score: 0.375 -> test with 'KNN' KNN tn, fp: 56559, 304 KNN fn, tp: 71, 25 KNN f1 score: 0.118 KNN cohens kappa score: 0.115 ====== Step 5/5 ======= -> Shuffling data -> Spliting data to slices ------ Step 5/5: Slice 1/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 56153, 710 GAN fn, tp: 24, 75 GAN f1 score: 0.170 GAN cohens kappa score: 0.167 -> test with 'LR' LR tn, fp: 53916, 2947 LR fn, tp: 16, 83 LR f1 score: 0.053 LR cohens kappa score: 0.050 LR average precision score: 0.612 -> test with 'GB' GB tn, fp: 56623, 240 GB fn, tp: 18, 81 GB f1 score: 0.386 GB cohens kappa score: 0.384 -> test with 'KNN' KNN tn, fp: 56702, 161 KNN fn, tp: 77, 22 KNN f1 score: 0.156 KNN cohens kappa score: 0.154 ------ Step 5/5: Slice 2/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 55608, 1255 GAN fn, tp: 10, 89 GAN f1 score: 0.123 GAN cohens kappa score: 0.121 -> test with 'LR' LR tn, fp: 55396, 1467 LR fn, tp: 6, 93 LR f1 score: 0.112 LR cohens kappa score: 0.109 LR average precision score: 0.781 -> test with 'GB' GB tn, fp: 56562, 301 GB fn, tp: 7, 92 GB f1 score: 0.374 GB cohens kappa score: 0.372 -> test with 'KNN' KNN tn, fp: 56379, 484 KNN fn, tp: 68, 31 KNN f1 score: 0.101 KNN cohens kappa score: 0.098 ------ Step 5/5: Slice 3/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 53055, 3808 GAN fn, tp: 12, 87 GAN f1 score: 0.044 GAN cohens kappa score: 0.040 -> test with 'LR' LR tn, fp: 55211, 1652 LR fn, tp: 12, 87 LR f1 score: 0.095 LR cohens kappa score: 0.092 LR average precision score: 0.656 -> test with 'GB' GB tn, fp: 56539, 324 GB fn, tp: 12, 87 GB f1 score: 0.341 GB cohens kappa score: 0.339 -> test with 'KNN' KNN tn, fp: 56461, 402 KNN fn, tp: 74, 25 KNN f1 score: 0.095 KNN cohens kappa score: 0.092 ------ Step 5/5: Slice 4/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227059 synthetic samples -> test with GAN.predict GAN tn, fp: 32505, 24358 GAN fn, tp: 1, 98 GAN f1 score: 0.008 GAN cohens kappa score: 0.005 -> test with 'LR' LR tn, fp: 54241, 2622 LR fn, tp: 11, 88 LR f1 score: 0.063 LR cohens kappa score: 0.060 LR average precision score: 0.745 -> test with 'GB' GB tn, fp: 56438, 425 GB fn, tp: 9, 90 GB f1 score: 0.293 GB cohens kappa score: 0.291 -> test with 'KNN' KNN tn, fp: 56668, 195 KNN fn, tp: 78, 21 KNN f1 score: 0.133 KNN cohens kappa score: 0.131 ------ Step 5/5: Slice 5/5 ------- -> Reset the GAN -> Train generator for synthetic samples -> create 227056 synthetic samples -> test with GAN.predict GAN tn, fp: 56149, 714 GAN fn, tp: 14, 82 GAN f1 score: 0.184 GAN cohens kappa score: 0.181 -> test with 'LR' LR tn, fp: 54754, 2109 LR fn, tp: 7, 89 LR f1 score: 0.078 LR cohens kappa score: 0.075 LR average precision score: 0.658 -> test with 'GB' GB tn, fp: 56468, 395 GB fn, tp: 8, 88 GB f1 score: 0.304 GB cohens kappa score: 0.302 -> test with 'KNN' KNN tn, fp: 56594, 269 KNN fn, tp: 72, 24 KNN f1 score: 0.123 KNN cohens kappa score: 0.121 ### Exercise is done. -----[ LR ]----- maximum: LR tn, fp: 55396, 3344 LR fn, tp: 16, 94 LR f1 score: 0.112 LR cohens kappa score: 0.109 LR average precision score: 0.854 average: LR tn, fp: 54682.24, 2180.76 LR fn, tp: 9.76, 88.64 LR f1 score: 0.079 LR cohens kappa score: 0.076 LR average precision score: 0.700 minimum: LR tn, fp: 53519, 1467 LR fn, tp: 5, 83 LR f1 score: 0.053 LR cohens kappa score: 0.049 LR average precision score: 0.568 -----[ GB ]----- maximum: GB tn, fp: 56653, 425 GB fn, tp: 19, 92 GB f1 score: 0.431 GB cohens kappa score: 0.430 average: GB tn, fp: 56534.52, 328.48 GB fn, tp: 11.68, 86.72 GB f1 score: 0.341 GB cohens kappa score: 0.339 minimum: GB tn, fp: 56438, 210 GB fn, tp: 7, 80 GB f1 score: 0.285 GB cohens kappa score: 0.283 -----[ KNN ]----- maximum: KNN tn, fp: 56702, 578 KNN fn, tp: 81, 33 KNN f1 score: 0.156 KNN cohens kappa score: 0.154 average: KNN tn, fp: 56521.36, 341.64 KNN fn, tp: 73.28, 25.12 KNN f1 score: 0.112 KNN cohens kappa score: 0.109 minimum: KNN tn, fp: 56285, 161 KNN fn, tp: 66, 18 KNN f1 score: 0.065 KNN cohens kappa score: 0.062 -----[ GAN ]----- maximum: GAN tn, fp: 56745, 56062 GAN fn, tp: 49, 99 GAN f1 score: 0.410 GAN cohens kappa score: 0.409 average: GAN tn, fp: 46575.64, 10287.36 GAN fn, tp: 17.0, 81.4 GAN f1 score: 0.136 GAN cohens kappa score: 0.133 minimum: GAN tn, fp: 801, 118 GAN fn, tp: 0, 50 GAN f1 score: 0.003 GAN cohens kappa score: 0.000