kNN of ML in action

[mimiao2017]

No description provided.

[mimiao2017]

the diff between shape(a) and a.shape:
shape(a):retry:a.shape,the same thing

tile

numpy.sum
http://blog.csdn.net/shengmingqijiquan/article/details/53009753

argsort

to be continued

[mimiao2017]

basics about np.array

1. when A is an array,A*B or multiply means elementwise product;
   A.dot(B)or dot(A,B) means martix product.
   when A is a matrix, multiply means elementwise product;
   A*B or A.dot(B) or dot(A,B) means martix product.
   when A ,B are of different types, multiply means elementwise product;
   A*B or A.dot(B) or dot(A,B) means martix product.

2. generate a certain size random array:np.random.random([2,3])

3. aoubt +=,TypeError: Cannot cast ufunc add output from dtype('float64') to dtype('int64') with casting rule 'same_kind'

4. upcasting:不同类型的数据进行计算，结果的类型取更通用/精确的类型。

5. 3D数组：创建一个3行4列2维数组：a = np.arange(24).reshape([2,3,4]).
   计算：axis=0，按维计算；axis=1，按列计算；axis=2，按行计算；
   2D数组：创建3行4列：a = np.arange(12).reshape([3,4]).
   计算：axis=0，按列计算；axis=1，按行计算；

6. b.cumsum(axis=1),按行累加

7.universal functions