512song/jupyter
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

20230206

Browse Source
This commit is contained in:
512song@sina.com committed 2023-02-26 11:24:18 +08:00
1 parent bffe1f6bed
commit a3e959f6f2
1 file changed
+131 -116
Download Patch File Download Diff File Expand all files Collapse all files
体测单位/北海炼化.ipynb
+131 -116
View File
@@ -10,27 +10,12 @@
},
{
"cell_type": "code",
"execution_count": 1,
"execution_count": null,
"id": "f6bc68b5-9ed8-4b28-ab1a-452107b6429a",
"metadata": {
"execution": {
"iopub.execute_input": "2022-11-15T06:00:04.369411Z",
"iopub.status.busy": "2022-11-15T06:00:04.368113Z",
"iopub.status.idle": "2022-11-15T06:00:11.722290Z",
"shell.execute_reply": "2022-11-15T06:00:11.719051Z",
"shell.execute_reply.started": "2022-11-15T06:00:04.369149Z"
},
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"ok\n"
]
}
],
"outputs": [],
"source": [
"import openpyxl\n",
"import json\n",
@@ -72,27 +57,12 @@
},
{
"cell_type": "code",
"execution_count": 17,
"execution_count": null,
"id": "97031317-76ed-4404-b922-6ca749e4fd2a",
"metadata": {
"execution": {
"iopub.execute_input": "2022-11-01T09:18:38.933088Z",
"iopub.status.busy": "2022-11-01T09:18:38.932552Z",
"iopub.status.idle": "2022-11-01T09:18:39.012745Z",
"shell.execute_reply": "2022-11-01T09:18:39.011599Z",
"shell.execute_reply.started": "2022-11-01T09:18:38.933039Z"
},
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"ok\n"
]
}
],
"outputs": [],
"source": [
"import json\n",
"import time\n",
@@ -147,16 +117,9 @@
},
{
"cell_type": "code",
"execution_count": 18,
"execution_count": null,
"id": "f70a02f2-934e-43ed-93ff-b34cb732dfc6",
"metadata": {
"execution": {
"iopub.execute_input": "2022-11-01T09:18:42.502192Z",
"iopub.status.busy": "2022-11-01T09:18:42.501603Z",
"iopub.status.idle": "2022-11-01T09:18:42.729776Z",
"shell.execute_reply": "2022-11-01T09:18:42.728671Z",
"shell.execute_reply.started": "2022-11-01T09:18:42.502143Z"
},
"tags": []
},
"outputs": [],
@@ -210,16 +173,9 @@
},
{
"cell_type": "code",
"execution_count": 4,
"execution_count": null,
"id": "7a0a9111-2f2c-4c5d-bb5e-3e4fbd7ab1aa",
"metadata": {
"execution": {
"iopub.execute_input": "2022-11-15T06:21:40.662548Z",
"iopub.status.busy": "2022-11-15T06:21:40.661365Z",
"iopub.status.idle": "2022-11-15T06:21:40.862211Z",
"shell.execute_reply": "2022-11-15T06:21:40.859943Z",
"shell.execute_reply.started": "2022-11-15T06:21:40.662431Z"
},
"tags": []
},
"outputs": [],
@@ -261,27 +217,12 @@
},
{
"cell_type": "code",
"execution_count": 24,
"execution_count": null,
"id": "c42c5c04-49f3-4b4c-8f77-e37cfa9748f2",
"metadata": {
"execution": {
"iopub.execute_input": "2022-11-01T09:27:40.105525Z",
"iopub.status.busy": "2022-11-01T09:27:40.104998Z",
"iopub.status.idle": "2022-11-01T09:27:40.151243Z",
"shell.execute_reply": "2022-11-01T09:27:40.150059Z",
"shell.execute_reply.started": "2022-11-01T09:27:40.105476Z"
},
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"ok!\n"
]
}
],
"outputs": [],
"source": [
"import json\n",
"import openpyxl\n",
@@ -331,27 +272,12 @@
},
{
"cell_type": "code",
"execution_count": 8,
"execution_count": null,
"id": "a9460ee7-1f60-4a3b-a45d-8a1dba04b920",
"metadata": {
"execution": {
"iopub.execute_input": "2022-10-31T11:54:40.324705Z",
"iopub.status.busy": "2022-10-31T11:54:40.324184Z",
"iopub.status.idle": "2022-10-31T11:54:40.626097Z",
"shell.execute_reply": "2022-10-31T11:54:40.624830Z",
"shell.execute_reply.started": "2022-10-31T11:54:40.324656Z"
},
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"ok\n"
]
}
],
"outputs": [],
"source": [
"import json\n",
"import openpyxl\n",
@@ -497,16 +423,9 @@
},
{
"cell_type": "code",
"execution_count": 8,
"execution_count": null,
"id": "26343c71-535f-445b-9274-6492febfd435",
"metadata": {
"execution": {
"iopub.execute_input": "2023-02-20T08:21:07.933556Z",
"iopub.status.busy": "2023-02-20T08:21:07.933029Z",
"iopub.status.idle": "2023-02-20T08:21:08.203245Z",
"shell.execute_reply": "2023-02-20T08:21:08.202135Z",
"shell.execute_reply.started": "2023-02-20T08:21:07.933507Z"
},
"tags": []
},
"outputs": [],
@@ -514,7 +433,7 @@
"import openpyxl\n",
"import json\n",
"\n",
"wb = openpyxl.load_workbook('data/北海体检报告整理数据2021.xlsx')\n",
"wb = openpyxl.load_workbook('data/中国石化北海炼化有限责任公司2022年健康体检团体报告统计表.xlsx')\n",
"sheet = wb.active\n",
"# sheets = wb.sheetnames\n",
"dict1 = {}\n",
@@ -531,10 +450,10 @@
" dict1[code].setdefault('bz',[]) \n",
" dict1[code]['bz'].append(sheet.cell(n, 6).value) \n",
" \n",
"#filename = 'data/北海炼化体检人员情况.json'\n",
"#with open(filename, 'w') as fl:\n",
"# json.dump(dict1, fl, ensure_ascii=False)\n",
"#print('ok')\n",
"filename = 'data/北海炼化2022年健康体检团体报告.json'\n",
"with open(filename, 'w') as fl:\n",
" json.dump(dict1, fl, ensure_ascii=False)\n",
"print('ok')\n",
"list1 = []\n",
"for k, v in dict1.items():\n",
" list2 = []\n",
@@ -546,7 +465,7 @@
" list2.append(','.join(mbz))\n",
" list1.append(list2)\n",
"title = ['体检编号','姓名','性别','年龄','体检症状']\n",
"filename = 'data/北海体检报告人员病症统计表2021年.xlsx'\n",
"filename = 'data/中国石化北海炼化有限责任公司2022年健康体检团体报告汇总统计表.xlsx'\n",
"wb = openpyxl.Workbook()\n",
"sheet = wb.active\n",
"sheet.append(title)\n",
@@ -566,27 +485,12 @@
},
{
"cell_type": "code",
"execution_count": 24,
"execution_count": null,
"id": "6f7b0837-a623-47fb-8451-3860df6ec133",
"metadata": {
"execution": {
"iopub.execute_input": "2023-02-24T06:12:40.001406Z",
"iopub.status.busy": "2023-02-24T06:12:40.000816Z",
"iopub.status.idle": "2023-02-24T06:12:40.212597Z",
"shell.execute_reply": "2023-02-24T06:12:40.211536Z",
"shell.execute_reply.started": "2023-02-24T06:12:40.001352Z"
},
"tags": []
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"[['2209190190', '李壮', '男', '26', '血脂异常'], ['2209190107', '何沛霖', '男', '33', '血脂异常'], ['2209190109', '唐六一', '男', '34', '血脂异常'], ['2209190149', '段龙飞', '男', '34', '血脂异常'], ['2209190139', '邵光明', '男', '35', '血脂异常'], ['2209190182', '杨友', '男', '35', '血脂异常'], ['2209190200', '连修志', '男', '35', '血脂异常'], ['2211090278', '张利雄', '男', '35', '血脂异常'], ['2209190188', '王广乾', '男', '37', '血脂异常'], ['2209190228', '张伟', '男', '37', '血脂异常'], ['2209190158', '唐良好', '男', '38', '血脂异常'], ['2209190160', '潘保民', '男', '38', '血脂异常'], ['2209190150', '张国剑', '男', '39', '血脂异常'], ['2209190124', '莫秋有', '男', '42', '血脂异常'], ['2209190212', '谢朝洲', '男', '42', '血脂异常'], ['2209190131', '王刚', '男', '44', '血脂异常'], ['2209190135', '周能', '男', '44', '血脂异常'], ['2209190108', '李文勇', '男', '45', '血脂异常'], ['2209190120', '郭兵', '男', '45', '血脂异常'], ['2209190163', '朱建喜', '男', '45', '血脂异常'], ['2209190180', '房浩', '男', '45', '血脂异常'], ['2209190093', '陈力平', '男', '46', '血脂异常'], ['2209190227', '朱传臻', '男', '46', '血脂异常'], ['2209190127', '吴容杰', '男', '47', '血脂异常'], ['2209190229', '杜青松', '男', '47', '血脂异常'], ['2209190191', '李建威', '男', '48', '血脂异常'], ['2209190100', '李卫军', '男', '50', '血脂异常'], ['2209190121', '李开锋', '男', '50', '血脂异常'], ['2209190123', '梁承李', '男', '50', '血脂异常'], ['2209190164', '田浩', '男', '50', '血脂异常'], ['2209190177', '阴雄才', '男', '50', '血脂异常'], ['2209190185', '郭小波', '男', '50', '血脂异常'], ['2209190095', '杜玉山', '男', '51', '血脂异常'], ['2209190198', '陈平', '男', '51', '血脂异常'], ['2209190199', '翟璟峰', '男', '51', '血脂异常'], ['2209190203', '赵学良', '男', '51', '血脂异常'], ['2209190099', '吴军', '男', '52', '血脂异常'], ['2209190133', '张远才', '男', '52', '血脂异常'], ['2209190157', '邱宁', '男', '52', '血脂异常'], ['2209190173', '丁少军', '男', '52', '血脂异常'], ['2209190192', '廖祁军', '男', '52', '血脂异常'], ['2209190196', '王健', '男', '52', '血脂异常'], ['2209190142', '黄葵东', '男', '53', '血脂异常'], ['2209190145', '伍荣滨', '男', '53', '血脂异常'], ['2209190168', '陈海文', '男', '53', '血脂异常'], ['2209190186', '王涛', '男', '53', '血脂异常'], ['2209190207', '杨作华', '男', '53', '血脂异常'], ['2209190103', '黄新友', '男', '54', '血脂异常'], ['2209190167', '莫有先', '男', '54', '血脂异常'], ['2209190189', '郝文化', '男', '54', '血脂异常'], ['2209190222', '范伟政', '男', '54', '血脂异常'], ['2209190226', '曾安友', '男', '54', '血脂异常'], ['2209190247', '陈伟', '男', '54', '血脂异常'], ['2209190175', '徐树友', '男', '55', '血脂异常'], ['2209190183', '吴文强', '男', '55', '血脂异常'], ['2209190240', '陈波', '男', '55', '血脂异常'], ['2209190242', '张惜光', '男', '55', '血脂异常'], ['2209190261', '陈彧', '男', '55', '血脂异常'], ['2209190187', '吴福强', '男', '56', '血脂异常'], ['2209190194', '贾春生', '男', '56', '血脂异常'], ['2209190089', '李继炳', '男', '57', '血脂异常'], ['2209190171', '陈华冠', '男', '57', '血脂异常'], ['2209190208', '吴冰', '男', '57', '血脂异常'], ['2209190258', '何国清', '男', '57', '血脂异常'], ['2209190090', '麦郁穗', '男', '58', '血脂异常'], ['2209190096', '谢涛', '男', '58', '血脂异常'], ['2209190262', '杜海惠', '男', '58', '血脂异常'], ['2209190102', '张宗荣', '男', '59', '血脂异常'], ['2209190255', '陈亚统', '男', '59', '血脂异常'], ['2209190259', '林祯海', '男', '59', '血脂异常'], ['2209190260', '吕钰', '男', '59', '血脂异常'], ['2209190130', '钟宇虹', '女', '26', '血脂异常'], ['2209150640', '蔡耐寒', '女', '33', '血脂异常'], ['2209190117', '李海娇', '女', '33', '血脂异常'], ['2209190236', '杨敏', '女', '34', '血脂异常'], ['2209190214', '邓施艳', '女', '36', '血脂异常'], ['2209190128', '代永香', '女', '40', '血脂异常'], ['2209190232', '黄蕊', '女', '43', '血脂异常'], ['2209190125', '何欣', '女', '46', '血脂异常'], ['2209190205', '梁艳玲', '女', '46', '血脂异常'], ['2209190216', '任志莹', '女', '46', '血脂异常'], ['2209190201', '甘志红', '女', '47', '血脂异常'], ['2209190122', '黄洁', '女', '49', '血脂异常'], ['2209190114', '赵智慧', '女', '52', '血脂异常'], ['2209190143', '徐业卫', '女', '52', '血脂异常'], ['2209190233', '刘科峰', '女', '52', '血脂异常'], ['2209190245', '杨萍', '女', '52', '血脂异常'], ['2209190115', 'Line truncated
]
}
],
"outputs": [],
"source": [
"import openpyxl\n",
"import json\n",
@@ -611,10 +515,121 @@
"wb.save(filename) \n"
]
},
{
"cell_type": "markdown",
"id": "1a05cd85-32a0-4c91-a281-9e7253562ba3",
"metadata": {},
"source": [
"### 北海体检情况统计"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "1815f0da-c80b-461a-a2cb-8d0382d2819f",
"id": "81fefc86-cd07-446c-ad08-bdb36b9e1722",
"metadata": {
"tags": []
},
"outputs": [],
"source": [
"import openpyxl\n",
"import json\n",
"import itertools\n",
"import operator\n",
"\n",
"filename = 'data/北海炼化2022年健康体检团体报告.json'\n",
"with open(filename,'r') as fl:\n",
" dict1 = json.load(fl)\n",
"bz = set()\n",
"for k, v in dict1.items():\n",
" for i in v['bz']:\n",
" bz.add(i)\n",
"print(len(bz))\n",
"dict2 = {}\n",
"dict3 = {}\n",
"i = 1\n",
"for n in [2,3,4]:\n",
" iter1 = itertools.combinations(bz, n)\n",
" list1 = list(iter1)\n",
" print(len(list1)) \n",
" \n",
" list2 = []\n",
" for item in list1:\n",
" \n",
" for k, v in dict1.items(): \n",
" list3 = []\n",
" if set(item).issubset(v['bz']):\n",
" dict3[i] =item \n",
" list3 = [k,i]\n",
" list2.append(list3)\n",
" i = i+1\n",
" dict2[n] = list2\n",
"dict2['bingzheng'] = dict3\n",
"filename = 'data/北海炼化2022年健康体检病症组合.json'\n",
"with open(filename,'w') as fl:\n",
" json.dump(dict2, fl, ensure_ascii=False) \n",
"print('ok')"
]
},
{
"cell_type": "markdown",
"id": "149104bd-76d8-4c42-b325-b0290439c3b2",
"metadata": {},
"source": [
"### 北海体检情况分析"
]
},
{
"cell_type": "code",
"execution_count": 46,
"id": "e1b2e465-3ebb-448f-a3df-e2aaab70e688",
"metadata": {
"execution": {
"iopub.execute_input": "2023-02-26T02:09:56.640500Z",
"iopub.status.busy": "2023-02-26T02:09:56.639970Z",
"iopub.status.idle": "2023-02-26T02:10:09.671647Z",
"shell.execute_reply": "2023-02-26T02:10:09.670363Z",
"shell.execute_reply.started": "2023-02-26T02:09:56.640450Z"
},
"tags": []
},
"outputs": [],
"source": [
"import openpyxl\n",
"import json\n",
"import itertools\n",
"import operator\n",
"\n",
"filename = 'data/北海炼化2022年健康体检病症组合.json'\n",
"with open(filename,'r') as fl:\n",
" dict1 = json.load(fl)\n",
"bz = dict1['bingzheng']\n",
"bz1 = dict1['2']\n",
"list1 = []\n",
"for k, v in bz.items():\n",
" list2 = []\n",
" i = 1\n",
" bzs = len(v)\n",
" for v1 in dict1[str(bzs)]:\n",
" if v1[1] == int(k):\n",
" \n",
" i+=1\n",
" list2 =[bzs,','.join(v),i]\n",
" list1.append(list2)\n",
"filename = 'data/中国石化北海炼化有限责任公司2022年病症统计表.xlsx'\n",
"wb = openpyxl.Workbook()\n",
"sheet = wb.active\n",
"#sheet.append(title)\n",
"for row in list1:\n",
" sheet.append(row)\n",
" \n",
"wb.save(filename) "
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "2dc319f7-ccca-469a-8db3-2ca82097ab8f",
"metadata": {},
"outputs": [],
"source": []