python聚类算法解决方案（rest接口/mpp数据库/json数据/下载图片及数据）

# -*- coding: utf-8 -*-

from flask import Flask, request, send_from_directory
from k_means import exec
import logging
app = Flask(__name__)

#1.服务器上更改为服务器地址，用于存放数据
dirpath = 'E:\\ruanjianlaowang'

#2. 测试连通性，软件老王
@app.route('/')
def index():
  return "Hello, World!"


#3. k-means算法 软件老王
@app.route('/getKmeansInfoByLaowang', methods=['POST'])
def getKmeansInfoByLaowang():
  try:
     result = exec(request.get_json(), dirpath)
  except IndexError as e:
    logging.error(str(e))
    return 'exception:' + str(e)
  except KeyError as e:
    logging.error(str(e))
    return 'exception:' + str(e)
  except ValueError as e:
    logging.error(str(e))
    return 'exception:' + str(e)
  except Exception as e:
    logging.error(str(e))
    return 'exception:' + str(e)
  else:
    return result

#4.文件下载（图片及csv）
@app.route("/<path:filename>")
def getImages(filename):
  return send_from_directory(dirpath, filename, as_attachment=True)

#5.启动
if __name__ == '__main__':
  app.run(host="0.0.0.0", port=5000, debug=True)

import pandas as pd
import dbgp as dbgp
from pandas.io import json
from numpy import *
import matplotlib.pyplot as plt
import numpy as np
plt.switch_backend('agg')
import logging

# 执行 软件老王
def exec(params, dirpath):

  #1.获取参数，软件老王
  sql = params.get("sql")
  xlines = params.get("xlines")
  ylines = params.get("ylines")
  xlinesname = params.get("xlinesname")
  ylinesname = params.get("ylinesname")
  grouplinesname = params.get("grouplinesname")

  times = int(params.get("times"))
  groupnum = int(params.get("groupnum"))
  url = params.get("url")
  name = params.get("name")

  #2. 校验是否为空，软件老王
  flag = checkparam(sql, xlines, ylines, times, groupnum)
  if not flag is None and len(flag) != 0:
    return flag

  #3. 从数据库获取数据，软件老王
  try:
    data = dbgp.queryGp(sql)
  except IndexError:
    return sql
  except KeyError:
    return sql
  except ValueError:
    return sql
  except Exception:
    return sql

  if data.empty:
    return "exception:此数据集无数据，请确认后重试"
  #4 调用第三方sklearn的KMeans聚类算法，软件老王
  # data_zs = 1.0 * (data - data.mean()) / data.std() 数据标准化，不需要标准话
  from sklearn.cluster import KMeans
  model = KMeans(n_clusters=groupnum, n_jobs=4, max_iter=times)
  model.fit(data) # 开始聚类

  return export(model, data, data, url, dirpath, name,grouplinesname,xlines, ylines,xlinesname,ylinesname)

  # 5.生成导出excel 软件老王
def export(model, data, data_zs, url, dirpath, name,grouplinesname,xlines, ylines,xlinesname,ylinesname):


  # #详细输出原始数据及其类别
  detail_data = pd.DataFrame().append(data)
  if not grouplinesname is None and len(grouplinesname) != 0:
    detail_data.columns = grouplinesname.split(',')

  r_detail_new = pd.concat([detail_data, pd.Series(model.labels_, index=detail_data.index)], axis=1) # 详细输出每个样本对应的类别
  r_detail_new.columns = list(detail_data.columns) + [u'聚类类别'] # 重命名表头
  outputfile = dirpath + name + '.csv'
  r_detail_new.to_csv(outputfile, encoding='utf_8_sig') # 保存结果


  #重命名表头
  r1 = pd.Series(model.labels_).value_counts() # 统计各个类别的数目
  r2 = pd.DataFrame(model.cluster_centers_) # 找出聚类中心
  r = pd.concat([r2, r1], axis=1) # 横向连接（0是纵向），得到聚类中心对应的类别下的数目
  r.columns = list(data.columns) + [u'类别数目'] # 重命名表头

  return generateimage(r, data_zs, url, dirpath, name,model,xlines, ylines,xlinesname,ylinesname)

#6.生成图片及返回json，软件老王
def generateimage(r, data_zs, url, dirpath, name,model,xlines, ylines,xlinesname,ylinesname):
  image = dirpath + name + '.jpg'

  #6.1 中文处理，软件老王
  plt.rcParams['font.sans-serif'] = ['simhei']
  plt.rcParams['font.family'] = 'sans-serif'
  plt.rcParams['axes.unicode_minus'] = False
  # 6.2 画图，生成图片，软件老王
  labels = model.labels_
  centers = model.cluster_centers_
  data_zs['label'] = labels
  data_zs['label'] = data_zs['label'].astype(np.int)
  # 图标集合
  markers = ['o', 's', '+', 'x', '^', 'v', '<', '>']
  colors = ['b', 'c', 'g', 'k', 'm', 'r', 'y']
  symbols = []
  for m in markers:
    for c in colors:
      symbols.append((m, c))
  # 画每个类别的散点及质心
  for i in range(0, len(centers)):
    df_i = data_zs.loc[data_zs['label'] == i]
    symbol = symbols[i]
    center = centers[i]

    x = df_i[xlines].values.tolist()
    y = df_i[ylines].values.tolist()

    plt.scatter(x, y, marker=symbol[0], color=symbol[1], s=10)
    plt.scatter(center[0], center[1], marker='*', color=symbol[1], s=50)

  plt.title(name)
  plt.xlabel(xlinesname)
  plt.ylabel(ylinesname)
  plt.savefig(image, dpi=150)
  plt.clf()
  plt.close(0)

  # 6.3 返回json数据给前端展示，软件老王
  result = {}
  result['image_url'] = url + '/' + name + '.jpg'
  result['details_url'] = url + '/' + name + '.csv'
  result['data'] = r[:200]  #显示200，多的话，相当于预览
  result = json.dumps(result, ensure_ascii=False)
  result = result.replace('\\', '')
  return result

def checkparam(sql, xlines, ylines, times, groupnum):
  if sql is None or sql.strip() == '' or len(sql.strip()) == 0:
    return "数据集或聚类数据列，不能为空"
  if xlines is None or xlines.strip() == '' or len(xlines.strip()) == 0:
    return "X轴，不能为空"
  if ylines is None or ylines.strip() == '' or len(ylines.strip()) == 0:
    return "Y轴，不能为空"
  if times is None or times <= 0:
    return "聚类个数，不能为空或小于等于0"
  if groupnum is None or groupnum <= 0:
    return "迭代次数，不能为空或小于等于0"

{"image_url":"http://10.192.168.1:5000/ruanjianlaowang_65652.jpg","details_url":"http://10.192.168.1:5000/ruanjianlaowang_65652.csv","data":{"empno":{"0":7747.2,"1":7699.625,"2":7839.0},"mgr":{"0":7729.8,"1":7745.25,"2":7566.0},"sal":{"0":2855.0,"1":1218.75,"2":5000.0},"comm":{"0":29.5110766,"1":117.383964625,"2":31.281453},"deptno":{"0":20.0,"1":25.0,"2":10.0},"类别数目":{"0":5,"1":8,"2":1}}}