{ "cells": [ { "cell_type": "markdown", "metadata": { "_cell_guid": "3b724374-790b-c01d-ba60-ba058ac795c6" }, "source": [ "We know the different ranges for Hounsfield Units and so come up with a few ranges that we then compare. The basic hypothesis is that **more blood means more cancer**" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "7f9f2235-1673-ea37-c484-0169660fee2c" }, "outputs": [], "source": [ "import os\n", "import numpy as np # linear algebra\n", "from dicom import read_file as read_dicom_image\n", "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n", "from glob import glob\n", "from itertools import groupby\n", "import matplotlib.pyplot as plt\n", "from skimage.color import label2rgb\n", "from skimage.segmentation import mark_boundaries" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "cd9e233d-d2be-98c5-3b13-1be404225721" }, "outputs": [], "source": [ "label_files = glob('../input/stage1_labels.csv')\n", "if len(label_files)>0:\n", " label_df = pd.read_csv(label_files[0])\n", " print(label_df.sample(3))\n", "else:\n", " from warnings import warn\n", " warn('Label file is missing')" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "1d2bb876-de03-ca1a-d99e-15049818681b" }, "outputs": [], "source": [ "# get all of the patients\n", "all_images = glob(os.path.join('..', 'input', 'sample_images', '*', '*'))\n", "folder_list = [(k, list(v)) for k, v in groupby(all_images, lambda x: os.path.split(os.path.dirname(x))[1])]" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "7278c5c9-7b04-2e13-1c44-6973b735276c" }, "outputs": [], "source": [ "n_img = read_dicom_image(all_images[150])\n", "_ = plt.hist(n_img.pixel_array.flatten(),np.linspace(-500, 500, 20))" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "42a5e0c4-4669-32a1-de71-b50658654c27" }, "outputs": [], "source": [ "# Calculate features for each image\n", "def show_area(in_dcm, min_val, max_val):\n", " in_img = in_dcm.pixel_array\n", " lab_img = (in_img>=min_val) \n", " lab_img &= (in_img<=max_val)\n", " in_img = ((in_img+1200.0)/1800.0).clip(0,1)\n", " return label2rgb(lab_img, image = in_img, bg_label = 0)\n", "def calc_area(in_dcm, min_val, max_val):\n", " pix_area = np.prod(in_dcm.PixelSpacing)\n", " return pix_area*np.sum((in_dcm.pixel_array>=min_val) & (in_dcm.pixel_array<=max_val))\n", "# Here are a list of simple features and ranges to look at in more detail\n", "feature_list = {\n", " 'blood': (30, 45),\n", " 'fat': (-100, -50),\n", " 'water': (-10, 10)\n", "}" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "b08c2f81-fd3c-bbe5-0cf5-009e980e1783" }, "outputs": [], "source": [ "plt.imshow(show_area(n_img, *feature_list['blood']))" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "6443614c-08f2-6fd6-8160-7f2218bb0495" }, "outputs": [], "source": [ "# since Kaggle doesn't have the label table in the ../input/ folder\n", "\n", "has_cancer = '0015ceb851d7251b8f399e39779d1e7d,006b96310a37b36cccb2ab48d10b49a3,008464bb8521d09a42985dd8add3d0d2,00edff4f51a893d80dae2d42a7f45ad1,0257df465d9e4150adef13303433ff1e,02801e3bbcc6966cb115a962012c35df,028996723faa7840bb57f57e28275e4c,04a8c47583142181728056310759dea1,05609fdb8fa0895ac8a9be373144dac7,059d8c14b2256a2ba4e38ac511700203,064366faa1a83fdcb18b2538f1717290,0708c00f6117ed977bbe1b462b56848c,07349deeea878c723317a1ce42cc7e58,07bca4290a2530091ce1d5f200d9d526,081f4a90f24ac33c14b61b97969b7f81,09d7c4a3e1076dcfcae2b0a563a28364,0acbebb8d463b4b9ca88cf38431aac69,0c0de3749d4fe175b7a5098b060982a1,0c37613214faddf8701ca41e6d43f56e,0c60f4b87afcb3e2dfa65abbbf3ef2f9,0d06d764d3c07572074d468b4cff954f,0f5ab1976a1b1ef1c2eb1d340b0ce9c4,0ff552aa083ecfabaf1cfd65b0a8e674,118be21b7e0c3058b29a524686391c66,11fe5426ef497bc490b9f1465f1fb25e,12e0e2036f61c8a52ee4471bf813c36a,13bb12b3b27d5a7b4b142503a1ae9e73,1427be78bcf4aba96c5054b697be9b5b,149cc798827099f8bdb97cf702027305,14f713c1ef037f6c531cffdff0e5fb2c,15aa585fb2d3018b295df8619f2d1cf7,169b5bde441e8aa3df2766e2e02cda08,184c61740244f4ce8fb985af9bb3d8e8,184fa4ae2b7ae010625d89f10186f1c5,185bc9d9fa3a58fea90778215c69d35b,198d3ff4979a9a89f78ac4b4a0fe0638,1dab3271160e1380c5a70a1e3ba40cb7,1e0f8048728717064645cb758eb89279,1f49f0c1d7feedcae9024d251797407c,1fb4887efd403cd9c0f6970fc8b679b5,229b8b785f880f61d8dad636c3dc2687,2365e0afe6844e955f3d4c23a16dc1a9,243e69389ae5738d3f89386b0efddbcd,245fe0c86269602b0dab44c345b0b412,2488c5b32e837dc848fe6fe4b1bbb7cb,2619ed1e4eca954af4dcbc4436ef8467,274a81c75d244187247789bd71de2b3a,281bb28a077ccfcd40ce4a543a5aea89,28a9b77a9113ce491433d3ea47fa8fc9,28e29fe26140703e5bbe570f982bd112,2969c7ad0e550fee1f4a68fcb3bbb9e5,29d92a1e253cef2c7f34c6db26ce11e3,2a20e4a4e6411f72374fdffebabfc235,2a2300103f80aadbfac57516d9a95365,2b861ff187c8ff2977d988f3d8b08d87,2c06f5c66f3c79515b7712605dea4400,2d5cd7c1ee9a74a1244ddd6b55ad0446,2d977650e6388d2c45825a77e94437a2,2e8bb42ed99b2bd1d9cd3ffaf5129e4c,2ebb1e8f14802c33f0e4215a7545d70d,2ed8eb4430bf40f5405495a5ec22a76d,2f154a687b94f7b59fec7048cbfb5354,2fc3d8ef26fc7aafad44d5034673dd4c,303b4b8425389134997a38b975c205d3,30b8aa7f5688cab5ff0964f34b715c4d,31136e50b7205e9184227f94cdea0090,318bf8045b625b40825552420abfe1ef,31f35f920a472a1c3eacb565fe027923,322bf0acacba9650fa5656b9613c75c8,3252220375d82c3720d36d757bb17345,3285ba0f447f3091c0c7c061b47c2f62,32cda856b7ec759fd3ebaa363c505e88,33dd6666d9f0338929ecce58bb7c4cc3,3457880b1a66030feb8adaed6da805af,348a53f500ada390ddd00cc47d310b2c,34c0760406297a3c8fd5077fb7cd95b0,352c23fe8a3d0640ea531a6bf223732c,35b9a3e9871499893f76c8e6c648562c,375a52b012066845a2eeb5032a92fc6b,380eb569a5750648434cc8ae8da4a0a9,383c27906392e9ce57f6ab5ef1cb6f62,385f1f49b0c20563177c36b7470f1c46,398208da1bcb6a88e11a7314065f13ff,39ebb8121ea6faec0405a4e8db883b55,3a5bbc2f1f5d6d76a48ba5300105d998,3f6431400c2a07a46386dba3929da45d,4001d754871a8da824b8444e32dc6e0f,40c95c9be0bd7c290534ce374c58bec9,437e42695e7ad0a1cb834e1b3e780516,43f2ef8f53e1aa03bfb6378d0c20a8ac,44988c6efa451e8d496188cb30669d44,4521c94debf37a4dc9f3b70366a21640,48e592418247393234dd658f9112c543,4a782bbc2608288a3ed05e511af6f8bb,4af17bcb31669a9eab0b6ef8e22a8dcf,4baa552f3a11782f39e16b345d401fb8,4cc8af2efef2f41bf70684be25276ce5,4cd70a98baca46b116071b32788d3c2d,4d7df08f074b221eec6311c2617a5ba8,4dbda61d574417c7f25d6e9a8f0749a7,4dcd34bd9b10f96453b63d4f55d1fd44,504e447ad62ea9ebb283873e044b5dd2,51bd5c556c77ecdaf489d8dd9f7a05f1,51fbac477a3639f983904fc4d42b8c15,5267ea7baf6332f29163064aecf6e443,54056288ab97cebc4b0ea33c23f47ff6,5572cc61f851b0d10d6fec913ae722b9,55c01868f1d9c37fa3f174dc3c0d44e8,56462ba8833fc842d16be7e311214404,570ea80b0dcc08f3e8751a6f4b2b1cd5,573a661e2d784f9385a3b78c9757ddad,57822feb6186b788c4e1877123428454,5782e6873c666529c6a66421acb043dc,592c2481f17d6a2cecfe7bbb6a27722c,59fc9d939f05bf3023c1387c1c086520,5ade88428e6463fa212d4c287228e8ed,5b412509bc40a3aeb3b5efef1fdfcfc9,5b642ed9150bac6cbd58a46ed8349afc,5c99ab7172afa78312fe73a3c0dd342f,5e0c8cba8eab51076ac0014049d770c1,5fd33ea74e1ad740a201ae9b3c383fc5,608202eb3c368512e55e9e339a203790,608a7028689c6ab3aea5f116007169b2,60b389fb2f7eeb912586d1a3ccc9dbbc,61630ec628631f7fe3980f869e1a4fbe,6171d57221e26d1f15d3c71fe966ab18,624a34fa8fd36847724e749877343847,627499714e279203bd1294290f8fc542,627836151c555187503dfe472fb15001,63b5be42543c98ac5392f1bfbda085bf,648c99653d512edc1d28dd8e7054ceab,64a5a866461a3b6006efb0075e04dffe,65073aadb60e398d8db1806f5ea2a082,6541df84fd779ba6513a530c128f4e9b,65a380c07d416f78e85545eaaa2916a1,662153a685fb4268361bfbaca5e9ca23,668bb968918c63fad7d65581825b1048,66a92d789e440d3dbef3c69d20e20694,676467220abd8e2104417c5213664ef9,678c5ec1360784e0fe797208069e0bbb,6799964c08ad5ce7740defcd3bd037a6,6857c76be618bb0ddced5f4fecc1695f,6969c031ee0c34053faff3aac9dd2da7,6be677ba1631174397b0c1e26a46af30,6cb2908fd789700db727dd96526bc342,6ee742b62985570a1f3a142eb7e49188,6f43af3f636f37b9695b58378f9265cc,6fd3af9174242c1b393fe4ba515e7a26,713d8136c360ad0f37d6e53b61a7891b,71665cc6a7ee85268ca1da69c94bbaeb,721949894f5309ed4975a67419230a3c,72fd04cf3099b148d9ad361efb988866,733205c5d0bbf19f5c761e0c023bf9a0,7395f64fba89c2463a1b13c400adf876,74b1b748971c474a8023f6406c54b18a,75aef267ad112a21c870b7e2893aaf8a,761aeadb65fb84c8d04978a75b2f684c,763ce10dfdd4662f15de3f5931d5534b,77033e4c1591403d1b1255607a20a983,775c5f8043e72b2284b5885254566271,77d6f5203d46073369d7038b2d58e320,7842c108866fccf9b1b56dca68fc355e,78c0a0104c0428e260cbd9e50eb7eea6,7917af5df3fe5577f912c5168c6307e3,799c0026d66479f7447ed0df5955f051,79e0e507b1cd1d0c8107de4fd6b9e444,7bfba4540956c0b2c5b78b3623a4855d,7c2b72f9e0f5649c22902292febdc89f,7d46ce019d79d13ee9ce8f18e010e71a,7fb1c8ffd78ca4b6869044251add36b4,80600d4a5fee7424d689ba7d0906d50f,817a99e1a60bcf4e37c904d73845ca50,820dd342da11af3a062d1647b3736fdd,823b5f08ce145f837066d2e19dab10c1,8298238a27be6111214a9bc711608181,8326bb56a429c54a744484423a9bd9b5,8369f716ca2d51c934e7f6d44cb156e9,839502f9ff68fd778b435255690f3061,84876a50f52476bcc2a63678257ae8b4,84a6c418d57bfc5214639012998356d4,85d59b470b927e825937ea3483571c6d,8601f5424bcf4cd8e7bc3d649e9995a2,868b024d9fa388b7ddab12ec1c06af38,87cdb87db24528fdb8479220a1854b83,87cdf4626079509e5d6d3c3b6c8bfc2e,880980cc7e88c83b0fea84f078b849e3,8815efa67adb15b2f8cfd49ec992f48e,882107a204c302e27628f85522baea49,88523579f4e325351665753e903cfdf5,88ae66cd575c45ec5bb0f1578e2f1c49,8918c484841c5d0a532fe146e9da61bd,893fbc465b9d8a25569659a2bac154ef,898bd4c517fb9cf94c7d06dae56b0136,89bfbba58ee5cd0e346cdd6ffd3fa3a3,8a97ff581c17a49a3ef97144efde8a19,8b6e16b4e1d1400452956578f8eb97c4,8c63c8ebd684911de92509a8a703d567,8e92c4db434da3b8d4e3cafce3f072fb,8ed68f2dbf103a4bc0fd8708d8c1ac93,8ee6f423ff988d10f2bb383df98c1b2e,90409f7fcfec3581033559f8340e48a9,90e3b396e1c1343a514eb5890833d3d8,90e5f4780b2f05136ff5f776a5cbc2af,91d29bc19205f8eb9a63de5b774a5575,92abfd85dd6afb639e9a8b60aaa08262,9397a41c9e819a92eb5c86e0e652d7c1,93a6f37a72f60498986374f57bfc30c4,95a27273c11db8bfb9fc27b1e64de6bd,9660e4a23b8dd7d5056a622ee3568a41,96acca47671874c41de6023942e10c16,9703fd051751879432975535663150da,9a3174ffe867f602ee82c512a01420ee,9b7524785a9bf40f0651deeb3b05b75f,9c779a4e5e56c77131f8e99d5eacb766,9de4a1ebcdf1cfd8566ed1d9b63cbeb7,9e5c2e760b94b8919691d344cfdbac7f,9e922147900b3984c9345bdda573e882,9e98136d07b953c3362e0a132c8810b6,9f52323d216f89d300612cfac0122d8b,a13d6c5f8f86d74e16c10cf9294bca31,a14e41eea93d7667a87d458d5cb28272,a162d204827e4e89a2e5ba81cc53247a,a19a122fe9a790576b57c6bd5cf9ff5c,a2d9e657a673798f9ebdfec1b361f93a,a32e7fdbc0db97e35aabd7c931a582ea,a4dc34f2731b62f60c6c390a39fe11b2,a532c6f9405e6f3a4229ea6e04b0d975,a70fd23bd8d535ffd42259cb91f4c5ca,a76b682e74918492c1f2ca4c13c29885,a784a51caee14229d46777f2a9770a5f,a853993fd839a0ee61f2ca73c4e497a6,a88c585e7d81744eec091a6f0600bd7b,a8e650f8494e894be06c9cee08702aa9,aa2747369e1a0c724bea611ea7e5ffcf,aa55708fcc8bf27b605bcd2fca0dc991,ac00af80df36484660203d5816d697aa,ac3345a5a05655c6bcce7d0b226a0042,ac366a2168a4d04509693b7e5bcf3cce,ac4c6d832509d4cee3c7ac93a9227075,ac57a379cfea05c07d9befe8b9359495,ac68eb0a3db3de247c26909db4c10569,ac9c16f3f287f0e0b321fb518ac71c75,ad7e6fe9d036ed070df718f95b212a10,adc3bbc63d40f8761c59be10f1e504c3,aea6f1621333074412b9a6acdcda31a9,af4dfdda000c16c4cb77ea236cf1e524,af6d573b8c6804e14e3a7b07a376e593,afb37b10bd304fa2c7b70cfaf1f489ed,b022a1d30d62ef2c1f0902f1a047a845,b158f44c31f4121c865c828ff79fc73d,b17cb533d71d63d548ce47b48b34c23c,b5de57869d863bdc1b84b0194e79a9d3,b635cda3e75b4b7238c18c6a5f1858f6,b6578699374a9954b9a8a9e7da2603b1,b6687898fe385b68d5ae341419ef3fdd,b6d8dd834f2ff1ed7a5154e658460699,b7045ebff6dbb0023087e0399d00b873,b7ef0e864365220b8c8bfb153012d09a,b83ce5267f3fd41c7029b4e56724cd08,b84c43bed6c51182d7536619b747343a,b8bb02d229361a623a4dc57aa0e5c485,b8dc33b670bb078d10954345c3ffbb3a,ba71b330a16e8b4c852f9a8730ee33b9,bb4b43d0dc4d9d2b61150df6556f6490,bbe21f027a1df4b07016b474b48d3f65,bc38f78d1194f57452f6bb5eed453137,bda661b08ad77afb79bd35664861cd62,be2be08151ef4d3aebd3ea4fcd5d364b,c020f5c28fc03aed3c125714f1c3cf2a,c05acf3bd59f5570783138c01b737c3d,c0625c79ef5b37e293b5753d20b00c89,c0f0eb84e70b19544943bed0ea6bd374,c1673993c070080c1d65aca6799c66f8,c1ba619e3b49e0cb7798bd10465c2b29,c2bdfb6ab5192656b397459648221918,c2e546795f1ea2bd7f89ab3b4d13e761,c3b05094939cc128a4593736f05eadec,c3e8db4f544e2d4ecb01c59551eb8ef0,c4c801ae039ba335fa32df7d84e4decb,c5887c21bafb90eb8534e1a632ff2754,c610439ebef643c7fd4b30de8088bb55,c67de8fbbe1e58b464334f93a1dd0447,c67e799bcc1e2635eb9164f6e8cf75f3,c8cfb917b0d619cb4e25f789db4641f8,cb64ff663195832e0b66a9bb17891954,cb94c3f894fc93c1ec0eb436c8564ed3,cd104ad99d5b939b6bdd28b154e28085,cd10ceca9862ba0cc2ffd0ed8c9b055c,cf0a772e90a14d77d664ce9baedf0e5c,cfbcb16fea277226d6771d8b1966397a,d09e4124b97b22ef45692b62b4ca7f03,d244870d213a21efa86e86c951d8c9a2,d2b47d9034d38a410f00dabba9754d91,d2eecd9f13a6d474338045d0c91cffbd,d6d5ed3055d084a6abf0f97af3fe2ff0,d7713d80767cfdaad5e3db537849d8d0,d777a77cc7a2ec2f1eed68799cc9075c,d7aa27d839b1ecb03dbf011af4bcb092,d7e5640b52c8e092ec277febc81478da,d81704ee56c124cc1434640918a3299c,d8ed783494996f55a587270a212f7d5b,d917c781760710015473eee9ce82e051,d92998a73d4654a442e6d6ba15bbb827,d991b1760fb8705de655a1da068f7a6a,d9a2bea7df4a888313374beb142cf9c0,da821546432756d377777d7f4c41ca2f,dbfbc12c7a943a2dc0e34bfd4a636bca,dc9854bcdcc71b690d9806438009001d,df015da931ad5312ee7b24b201b67478,df761dd787bfc439890740ccce934f36,e00832e96709eb85f8e0e608ca02c2b5,e10c2b829c39d4a500c09caf04d461a1,e2b7fe7fbb002029640c0e65e3051888,e3423505ef6b43f03c5d7bde52a5a78c,e38789c5eabb3005bfb82a5298055ba0,e43afa905c8e279f818b2d5104f6762b,e537c91cdfa97d20a39df7ef04a52570,e54b574a7e7c650edc224cbdede9e675,e56b9f25a47a42f4ae4085005c46109c,e572e978c2b50aca781e6302937e5b13,e58b78dc31d80a50285816f4ecd661e3,e5c68cfa0f33540da3098800f0daae2c,e5cf847e616cc2fe94816ffa547d2614,e659f6517c4df17e86d4d87181396ea6,e6b3e750c6c7a70ca512d77defcfe615,e709901da9ba15a95d4a29906edc01dd,e8eb842ee04bbad407f85fe671f24d4f,e9ccf1ce85c39779fafb9ec703c71555,ea7373271a2441b5864df2053c0f5c3e,eaf753dc137e12fd06e96d27f3111043,eb008af181f3791fdce2376cf4773733,ebd601d40a18634b100c92e7db39f585,ed0f3c1619b2becec76ba5df66e1ea56,ed49b57854f5580658fb3510676e03dd,eed4db0cb0576c274de569e98a56a270,f17867cc3e579dc2fc6f0334bc43a91d,f1a64fda219db48bcfb8ad3823ef9fc1,f25c425c827b35fcbaa23f2ed671540b,f29d00ddf6d9846aa600c3f0edf5f952,f2ca85bb9ae82a3d79b9f321f727ac19,f2f23b265b2a3b977cb81fe3193d7c2c,f3eafe72b1e9528116f3c430ab73a2ae,f3f6f40ccb01276d722d52701cab1754,f42a0343e5b5154c6a184fc955d8f20f,f467795ce3b50a771085d79ae8d29ecc,f5717f7cbc08d8bd942cd4c1128e3339,f63f2f63e2619012b4c798fd638c8b8a,f6c9e875d7adfe7add08f43528810f72,f725f46908f16062fd12c141eb47c6a7,f76143416ee2c8e1251f45f108fed468,f7a03adba817f2a2249b9dee0586f4be,f8f66fca04d2e67eacd86ea154827a4c,f938f9022abf7f1072fe9df79db7eccd,fa45178d023325b255a3d4fc3e96cb7d,fb57fc6377fd37bb5d42756c2736586c,fb99a80cbb2f441bb90135bab5b029fe,fbae4d04285789dfa32124c86586dd09,fc545aa2f58509dc6d81ef02130b6906,fd0c2dfe0b0c58330675c3191cef0d5b,fd2dd970bd3d91e5b26d7e57c03f70af,fda187bfb1d6a2ecd4abd862c7f7f94c,fe45462987bacc32dbc7126119999392'.split(',')\n" ] }, { "cell_type": "markdown", "metadata": { "_cell_guid": "255cc32a-e440-31b2-3f66-ab9a05e7eef2" }, "source": [ "# Simple Classification\n", "Now that we have a simple table we can try examining simple features to get a better feeling for the data on the sample images" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "713dd9ae-7401-b10d-2e26-e248a39a16b1" }, "outputs": [], "source": [ "simple_features = [{'patient': c_folder, 'cancer': c_folder in has_cancer, \n", " 'slice_count':len(c_files)} for c_folder, c_files in folder_list]" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "b11084be-24d5-ef79-fd79-8f0f9b284b73" }, "outputs": [], "source": [ "simple_df = pd.DataFrame(simple_features)\n", "simple_df.sample(3)" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "00409863-ece2-c72d-665b-ff021701e42a" }, "outputs": [], "source": [ "simple_df.plot.scatter('slice_count', 'cancer')" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "ff9d5d29-b0d2-5aee-dc1c-f3d47f5b417a" }, "outputs": [], "source": [ "blood_features = [{'patient': c_folder, 'cancer': c_folder in has_cancer, \n", " 'blood_in_first_slice': calc_area(read_dicom_image(list(c_files)[0]), *feature_list['blood'])} for c_folder, c_files in folder_list]\n", "bf_df = pd.DataFrame(blood_features)\n", "bf_df.sample(5)" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "11fade09-00d2-71b6-6a2a-545d3677c476" }, "outputs": [], "source": [ "bf_df.plot.scatter('blood_in_first_slice', 'cancer')" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "90ebbc23-e0bd-bb19-585d-1d5d3ce71810" }, "outputs": [], "source": [ "bf_df.plot.hist('blood_in_first_slice')" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "11b52de6-9ef4-1f43-2947-77e41629717c" }, "outputs": [], "source": [ "from sklearn.model_selection import train_test_split\n", "X_train, X_test, y_train, y_test = train_test_split(bf_df[['blood_in_first_slice']], \n", " bf_df.cancer, \n", " random_state = 12345,\n", " train_size = 0.8,\n", " stratify = bf_df.cancer)\n", "print('Training patients:{}, testing patients:{}'.format(X_train.shape[0], X_test.shape[0]))" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "d0e74258-1d4e-44ef-ea37-6dfceb76df21" }, "outputs": [], "source": [ "import sklearn.tree\n", "simple_dt = sklearn.tree.DecisionTreeClassifier()" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "b271397f-b4f5-fe70-5dd8-47aa848fcbad" }, "outputs": [], "source": [ "simple_dt.fit(X_train, y_train)" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "6f0daa04-462a-a4de-ffea-bb694f64f5ae" }, "outputs": [], "source": [ "X_pred = simple_dt.predict(X_test)" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "49ebd9c1-08d7-5681-c18d-dfde5c04c142" }, "outputs": [], "source": [ "from sklearn.metrics import confusion_matrix\n", "conf_mat = confusion_matrix(y_test, X_pred)\n", "plt.matshow(conf_mat, cmap = 'bone')\n", "print(conf_mat)" ] }, { "cell_type": "code", "execution_count": null, "metadata": { "_cell_guid": "ae507697-9250-fdd0-589c-3e8b535d2ca8" }, "outputs": [], "source": [ "# now show the result for all samples\n", "All_pred = simple_dt.predict(bf_df[['blood_in_first_slice']])\n", "all_conf_mat = confusion_matrix(bf_df['cancer'], All_pred)\n", "\n", "plt.matshow(all_conf_mat, cmap = 'bone')\n", "print(all_conf_mat)" ] } ], "metadata": { "_change_revision": 0, "_is_fork": false, "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.6.0" } }, "nbformat": 4, "nbformat_minor": 0 }