data and dream

1 用通俗的语言介绍下线性回归->逻辑回归->SVM之间的区别和联系。

2 聚类算法的应用场景，以及k-means中的k值怎么确定。

 def center(data):

     center = []
     for num in data:
         sumX = 0; sumY = 0
         for j in num:
             sumX += j[0]
             sumY += j[1]
             x = float(sumX) / len(data)
             y = float(sumY) / len(data)
         center.append([x, y])

     return center

 def distance(one, two):

     sumT = 0
     for i in range(len(one)):
         sumT += pow((one[i] - two[i]), 2)

     return pow(sumT, 0.5)

 def update(data, kcenter):

     length = len(kcenter)
     ret = [0] * length
     for i in range(length):
         ret[i] = []

     for num in data:
         tmp = []
         for point in kcenter:
            tmp.append(distance(num, point))
         ret[tmp.index(min(tmp))].append(num)

     return ret

 if __name__ == '__main__': 

     data = [(1, 2), (2, 3), (1, 6), (8, 9)]
     kcenter = [[0.2, 1.2], [2, 3]]
     error = 0.0000001

     while True:
       rt = update(data, kcenter)
       tmp = center(rt)
       sume = 0
       for sa in range(len(kcenter)):
          sume += distance(tmp[sa], kcenter[sa])
       if sume < error:
           print rt
           break
       else:
           kcenter = tmp
      

Kmeans

3 协同过滤中评分矩阵中的元素怎么确定。大矩阵怎么分解。

4 文本挖掘怎么处理。