python学习笔记

NUMPY：

import numpy as np

a = np.arange(20)，np.random.rand(5)，np.linspace(0, 2, 9)

a = a.reshape(4, 5)

type(a)

a.ndim，a.shape，a.size，a.dtype

d = (4, 5)，np.zeros(d)，np.ones(d, dtype=int)

raw = [[0,1,2,3,4], [5,6,7,8,9]]

b = np.array(raw)

a = np.asmatrix(a),b = np.mat(b)

b = np.matrix('1.0 2.0; 3.0 4.0')

全部的'+'，'-'，'*'，'/'运算都是基于全部的数组元素的

'+='、'-='、'*='、'/='操作符在NumPy中同样支持：

print np.exp(a)

print np.sqrt(a)

print np.square(a)

print np.power(a, 3)

a = np.arange(20).reshape(4,5)

print "a:"

print a

print "sum of all elements in a: " + str(a.sum())

print "maximum element in a: " + str(a.max())

print "minimum element in a: " + str(a.min())

print "maximum element in each row of a: " + str(a.max(axis=1))

print "minimum element in each column of a: " + str(a.min(axis=0))

stock_prices['kdj_j_yest']=np.insert(np.array(stock_prices['kdj_j'][:-1]),0,None)

布尔操作：由于Python中的布尔运算使用and、or和not等关键字，它们无法被重载，因此数组的布尔运算只能通过相应的ufunc函数进行。这些函数名都以“logical_”开头，np.logical_and np.logical_not np.logical_or  np.logical_xor

>>> a == b

array([False, False,  True, False, False], dtype=bool)

>>> a > b

array([False, False, False,  True,  True], dtype=bool)

>>> np.logical_or(a==b, a>b) # 和 a>=b 相同

print a[0][1]

print a[0, 1]

b = a.copy()

a[:,[1,3]]

a[:, 2][a[:, 0] > 5]

np.nan_to_num(a)

c = np.hstack([a,b])

d = np.vstack([a,b])

serices 的创建：

s = Series(np.random.randn(5), index=['a', 'b', 'c', 'd', 'e'], name='my_series')

d = {'a': 0., 'b': 1, 'c': 2}，s = Series(d)

Series(d, index=['b', 'c', 'd', 'a'])，Series(4., index=['a', 'b', 'c', 'd', 'e'])

serices 的选择和boolean过滤：

s[:2]，s[[2,0,4]]，s[['e', 'i']]

s[s > 0.5]

'e' in s

DataFrame的创建：

df = DataFrame(columns=('lib', 'qty1', 'qty2'))或df = DataFrame(d, index=['r', 'd', 'a'], columns=['two', 'three'])

d = {'one': Series([1., 2., 3.], index=['a', 'b', 'c']), 'two': Series([1., 2., 3., 4.], index=['a', 'b', 'c', 'd'])}，df = DataFrame(d)

d = {'one': [1., 2., 3., 4.], 'two': [4., 3., 2., 1.]}，df = DataFrame(d, index=['a', 'b', 'c', 'd'])

d= [{'a': 1.6, 'b': 2}, {'a': 3, 'b': 6, 'c': 9}]，df = DataFrame(d)

行拼接：for i in range(5):a = DataFrame([np.linspace(i, 5*i, 5)], index=[index[i]])，df = pd.concat([df, a], axis=0)

列拼接：a = Series(range(5))，b = Series(np.linspace(4, 20, 5))，df = pd.concat([a, b], axis=1)

(或pd.merge(df1,df2,left_index=True,right_index=True,how='outer'))

列扩展：val = Series([-1.2, -1.5, -1.7], index=['two', 'four', 'five'])，frame2['debt'] = val

行扩展：  df.loc[len(df)] = row或者res=res.append(a new pd.Series),

删除行、列：del DF['column-name']或者df.drop([Column Name or list],inplace=True,axis=1)（凡是会对原数组作出修改并返回一个新数组的，往往都有一个 inplace可选参数。如果手动设定为True（默认为False），那么原数组直接就被替换。也就是说，采用inplace=True之后，原数组名（如2和3情况所示）对应的内存值直接改变；而采用inplace=False之后，原数组名对应的内存值并不改变，需要将新的结果赋给一个新的数组或者覆盖原数组的内存位置）

DataFrame属性

print df.index

print df.values

DataFrame选择和切片：

目前初步确立用法：[][]时列行，[,]时行列，iloc，loc指定的是行

print df['b'][2]

print df['b']['gamma']

print df.iloc[1]

print df.loc['beta']

print df[1:3]

bool_vec = [True, False, True, True, False]

print "Selecting by boolean vector:"

print df[bool_vec]

print df[['b', 'd']].iloc[[1, 3]]

print df.iloc[[1, 3]][['b', 'd']]

print df[['b', 'd']].loc[['beta', 'delta']]

print df.loc[['beta', 'delta']][['b', 'd']]

最快访问：at,iat

print df.iat[2, 3]，print df.at['gamma', 'd']

最快访问的智能化：ix

print df.ix['gamma', 4]，print df.ix[['delta', 'gamma'], [1, 4]]，print df.ix[[1, 2], ['b', 'e']]

索引扩展

df = df.reindex(df.index|set(['e']))

df = df.reindex(list(df.index).append( [ 'c', 'd', 'e' ]))

pd.set_option('display.width', 200)

dates = pd.date_range('20150101', periods=5)

df = pd.DataFrame(np.random.randn(5, 4),index=dates,columns=list('ABCD'))

df2 = pd.DataFrame({ 'A' : 1., 'B': pd.Timestamp('20150214'), 'C': pd.Series(1.6,index=list(range(4)),dtype='float64'), 'D' : np.array([4] * 4, dtype='int64'), 'E' : 'hello pandas!' })

stock_list = ['000001.XSHE', '000002.XSHE', '000568.XSHE', '000625.XSHE', '000768.XSHE', '600028.XSHG', '600030.XSHG', '601111.XSHG', '601390.XSHG', '601998.XSHG']

raw_data = DataAPI.MktEqudGet(secID=stock_list, beginDate='20150101', endDate='20150131', pandas='1')

df = raw_data[['secID', 'tradeDate', 'secShortName', 'openPrice', 'highestPrice', 'lowestPrice', 'closePrice', 'turnoverVol']]

print df.shape

print df.head()

print df.tail(3)

print df.describe()

print df.sort(columns='tradeDate').head()

df = df.sort(columns=['tradeDate', 'secID'], ascending=[False, True])

print df[df.closePrice > df.closePrice.mean()].head()

print df[df['secID'].isin(['601628.XSHG', '000001.XSHE', '600030.XSHG'])].head()

print df.dropna(subset=['closePrice']).shape

print df.dropna(thresh=6).shape

print df.dropna(how='all').shape

print df.dropna().shape

print df.fillna(value=20150101).head()

print df['closePrice'].value_counts().head()

print df[['closePrice']].apply(lambda x: (x - x.min()) / (x.max() - x.min())).head()

dat1 = df[['secID', 'tradeDate', 'closePrice']].head()

dat2 = df[['secID', 'tradeDate', 'closePrice']].iloc[2]

dat = dat1.append(dat2, ignore_index=True)

dat1 = df[['secID', 'tradeDate', 'closePrice']]

dat2 = df[['secID', 'tradeDate', 'turnoverVol']]

dat = dat1.merge(dat2, on=['secID', 'tradeDate'])

df_grp = df.groupby('secID')

grp_mean = df_grp.mean()

df2 = df.sort(columns=['secID', 'tradeDate'], ascending=[True, False])

print df2.drop_duplicates(subset='secID')

print df2.drop_duplicates(subset='secID', take_last=True)

dat = df[df['secID'] == '600028.XSHG'].set_index('tradeDate')['closePrice']

dat.plot(title="Close Price of SINOPEC (600028) during Jan, 2015")

obj.combine_first(other) 方法的作用是使用 other 中的数据去填补 obj 中的 NA 值，就像打补丁。而且可以自动对齐。

marketIndex[field]=(marketIndex[field]+stock_prices[field]).combine_first(marketIndex[field])

pythonMap操作：

dict["w"] = "watermelon"

dict.get("c", "apple") 

print dict.pop("b")

dict.update(dict2)

dict.items()

dict.keys()

dict.values()

dict.clear()

print d.popitem()

#按照key排序print sorted(dict.items(), key=lambda d: d[0])

#按照value排序print sorted(dict.items(), key=lambda d: d[1])

#字典的浅拷贝dict2 = dict.copy()

#深拷贝import copy，dict3 = copy.copy(dict)

Python两列表同位置同元素个数

a:[0, 0, 1, 0, 0, 0, 0, 0, 0, 1]

b:[1, 0, 1, 1, 0, 0, 0, 0, 0, 0]

map(cmp,a,b).count(0)

若是只判断“1” 的个数呢

map(lambda x,y:x+y, a, b).count(2)

或者sum(map(lambda x, y: 1 if x==y==1 else 0, a, b))

参考：https://uqer.io/community/share/54ca15f9f9f06c276f651a56