对于python进行数据处理来说,pandas式一个不得不用的包,它比numpy很为强大。通过对《利用python进行数据分析》这本书中介绍pandas包的学习,再加以自己的理解,写下这篇随笔,与一起喜欢数据分析的朋友分享和相互学习。
import numpy as np import pandas as pd from pandas import Series, DataFrame # 函数反应和映射 df = DataFrame(np.random.randn(4,3), columns= list("bde"), index= ["Utah", "Ohio", "Texas", "Oregon"]) # print df # print np.abs(df) # 将函数应用到各列或行所形成的一维数组上。 f = lambda x : x.max() - x.min() # 每一列的最大值减最小值 # print df.apply(f, axis=0) # 每一行的最大值减最小值 # print df.apply(f, axis=1) # 返回值由多个值组成的Series def f(x): return Series([x.min(), x.max()], index=["min","max"]) # print df.apply(f) # 保留两位小数点 format = lambda x : "%.2f" % x # print df.applymap(format) # print df["e"].map(format) # 排序和排名 obj = Series(np.arange(4.), index=["b","a","d","c"]) # print obj.sort_index() frame = DataFrame(np.arange(8).reshape((2,4)),index=["three","one"], columns=["d",'a','b','c']) # 按照索引的行进行排序 # print frame.sort_index(axis=1) # 按照索引的列进行排序 # print frame.sort_index(axis=0) # 按照值的列进行排序(必须传入一个列的索引且只能排列一组) # print frame.sort_values('b', axis=0, ascending=False) # 按照值的行进行排序(必须传入一个行的索引且只能排列一组) # print frame.sort_values("one", axis=1, ascending=False) # 根据多个列进行排序 # print frame.sort_index(by=["a","b"]) # 排名 obj1 = Series([7,-5,7,4,2,0,4]) # print obj1.rank() # 加减乘除 add代表加,sub代表减, div代表除法, mul代表乘法 df1 = DataFrame(np.arange(12).reshape((3,4)), columns=list("abcd")) df2 = DataFrame(np.arange(20).reshape((4,5)), columns=list("abcde")) # print df1 + df2 # 将缺失值用0代替 # print df1.add(df2, fill_value=0) # 再进行重新索引时,也可以指定一个填充值 # print df1.reindex(columns=df2.columns, fill_value=0) data = {"state": ["Ohio","Ohio","Ohio","Nevada","Nevada"], "year" : [2000, 2001, 2002, 2001, 2002], "pop" : [1.5, 1.7, 3.6, 2.4, 2.9]} frame = DataFrame(data) # print frame # 矩阵的横坐标 # print frame.columns # 矩阵的纵坐标 # print frame.index # 获取列通过类似字典标记的方式或属性的方式,可以将DataFrame的列获取为一个Series: # print frame["state"] # print frame.year # 获取行也通过类似字典标记的方式或属性的方式,比如用索引字段ix # print frame.ix[3] # 修改列的内容 frame["debt"] = 16.5 # print frame # 精准匹配 val = Series([-1.2, -1.5, -1.7], index=["two", "four", "five"]) frame.index = Series(['one', 'two', 'three', 'four', 'five']) frame.debt = val # print frame # 为不存在的列赋值存在列中的某个值会创建出一个布尔列。关键字del用于删除列。 frame["eastern"] = frame.state == "Ohio" # print frame del frame["eastern"] # 只能这样表示 # print frame # 嵌套字典 pop = { "Nevada" : {2001 : 2.4, 2002 : 2.9}, "Ohio" : {2000 : 1.5, 2001 : 1.7, 2002 : 3.6} } # 传给DataFrame,它会被解释为:外层字典的键作为列,内层键则作为行索引 frame2 = DataFrame(pop) # print frame2 # 对该结果进行转置 # print frame2.T # 内层字典的键会被合并、排序以形成最终的索引。 frame3 = DataFrame(pop, index=[2001, 2002, 2003]) # print frame3 frame3.index.name = "year"; frame3.columns.name = "state" # print frame3 # 重新索引 obj = Series([4.5, 7.2, -5.3, 3.6], index=["d", "b", "a", "c"]) # reindex将会根据新索引进行重排。 obj2 = obj.reindex(["a", "b", "c", "d", "e"]) # print obj2 # 将缺失值用0代替 obj2 = obj.reindex(["a", "b", "c", "d", "e"], fill_value= 0) # print obj2 # 插值处理--Series obj3 = Series(["blue", "purple", "yellow"], index=[0,2,4]) # 前向填充ffill或pad a = obj3.reindex(xrange(6), method="ffill") # print a # 后向填充bfill或backfill b = obj3.reindex(xrange(6), method="bfill") # print b # 插值处理--DataFrame import numpy as np f = DataFrame(np.arange(9).reshape((3,3)), index=["a","c","d"], columns=["Ohio", "Texas", "California"]) # 改变行的索引 f2 = f.reindex(["a","b","c","d"], fill_value=9) # print f2 # 改变列的索引 col = ["Texas", "Utah", "California"] f3 = f.reindex(columns=col) # print f3 # 同时改变列和行的索引 f4 = f.reindex(["a","b","c","d"], method="ffill", columns=["Texas", "Utah", "California"]) # print f4 # 丢弃指定轴上的项--Series mys = Series(np.arange(5.), index=["a","b","c","d","e"]) # print mys # drop()删除某个索引以及对应的值 mys_new = mys.drop("c") # print mys_new mys_new1 = mys.drop(["c","e"]) # print mys_new1 # 丢弃指定轴上的项--DataFrame data = DataFrame(np.arange(16).reshape((4,4)), index=["Ohio", "Colorado", "Utah", "New York"], columns=["one", "two", "three", "four"]) # 删除某行轴上的值 data1 = data.drop(["Ohio","Utah"], axis=0) # axis=0代表行 # print data1 # 删除某列轴上的值 data2 = data.drop(["one","three"], axis=1) # axis=1代表列 # print data2 obj = Series(range(5), index=['a', 'a', 'b', 'b', 'c']) # 使用is_unique属性可以知道他的值是否是唯一的 print obj.index.is_unique # obj['a'] df = DataFrame(np.random.randn(4, 3), index=['a', 'b', 'a', 'b']) print df.ix["b", 1] print df[1]
pandas中的索引高级处理:
from pandas import Series, DataFrame import pandas as pd import numpy as np # 索引、选取和过滤--Series obj = Series(np.arange(4), index=["a","b","c","d"]) # print obj["b"] # print obj[1] # print obj[2:4] # print obj[["b","a","d"]] # print obj[[1,3]] # print obj[obj < 2] # 利用标签的切片运算与普通的python切片运算不同,其末端是包含的 # print obj["b":"c"] obj["b":"c"] = 5 # print obj # 索引、选取和过滤--DataFrame data = DataFrame(np.arange(16).reshape((4, 4)), index=["Ohio", "Colorado", "Utah", "New York"], columns=["one", "two", "three", "four"]) # 选取某列的值 # print data["two"] # 选取多个列的值 # print data[["two","one"]] # 通过切片或布尔型数组选取行的值 a = data[:2] b = data[data["three"] > 5] # data[data < 5] = 0 # print data # 选取出列和行的值,用ix[行,列] c = data.ix["Ohio","two"] # print c, data # print data.ix["Ohio",["two","three"]] # 可以用数字代替列的轴 # print data.ix[["Ohio","Colorado"],[3,0,1]] # 也可以用数字代替行的轴 # print data.ix[[0,1],[3,0,1]] d = data.ix[:"Utah", "two"] # 行中每个值大于5且前三列的值 e = data.ix[data.three > 5, :3] # print e # Series的字符串表现形式为:索引在左边,值在右边。如果没有指定索引,那么默认从0到(N-1)的整数型索引。 # 可以通过values和index属性获取数组的形式和索引。 obj = Series([2,3,-6,7]) # print obj # print obj.values # print obj.index obj2 = Series([2,3,-6,7],index=["d","b","a","c"]) # print obj2.index # print obj2["a"] obj2["d"] = 6 # print obj2[["c","a","d"]] # print obj2[obj2 > 0 ] # print obj2 * 2 # print np.exp(obj2) sdata = {"Ohio" : 35000, "Texas" : 71000, "Oregon" : 16000, "Utah" : 5000} # 直接用字典建立数组 obj3 = Series(sdata) # 如果只传入一个字典,则结果Series中的索引就是原字典的键。 states = ["California","Ohio","Oregon","Texas"] obj4 = Series(sdata, index=states) # 上述obj4中California在对应的sdata中找不到对应值,所以用NaN表示(缺失值) # 检测是否有缺失值。 pd.isnull(obj4) pd.notnull(obj4) obj4.isnull() # Series最重要的一个功能是:它在算术运算中会自动对齐不同的索引的数据。 # print obj3 + obj4 # Series对象和索引都有一个name属性,该属性跟pandas其他的关键功能关系非常密切: obj4.name = "population" obj4.index.name = "state" # print obj4 # Series的索引可以通过赋值的方式就地修改 obj.index = ["Bob","Steve","Jeff","Ryan"] print obj
用pandas包进行简单的统计学计算:
import numpy as np import pandas as pd from pandas import Series, DataFrame df = DataFrame([[1.4, np.nan], [7.1, -4.5], [np.nan, np.nan],[0.75, -1.3]], index=['a','b','c','d'], columns=["one","two"]) # print df.sum() # 传入axis=1将会按行进行求和运算 # print df.sum(axis=1) # NA值会自动被排除,除非整个切片是NA值。可以通过skipna选项禁止这种功能 d = df.mean(axis=1, skipna=False) f = lambda x : "%.2f" % x # print d.apply(f) # 统计 # 间接统计 # print df.idxmax() # 累计型统计(前一项加后一项) # print df.cumsum() # 一次性汇总统计 # print df.describe() # print df.min(axis=1) # 计算相关系数和协方差 obj = DataFrame(np.random.randn(5,4), index=["2009-12-24","2009-12-28","2009-12-29","2009-12-30","2009-12-31"], columns=["AAPL","GOOG","IBM","MSFT"]) obj.index.name = "Data" # print obj # index 代表行, columns 代表列 # corr方法用于计算两个Series中重叠的、非NA的、按索引对齐的值的相对系数。cov用于计算协方差: # print obj.MSFT.corr(obj.IBM) # print obj.MSFT.cov(obj.IBM) # 用于DataFrame的corr和cov # 相关系数 # print obj.corr() # 协方差 # print obj.cov() # 按列或行跟一个Series或Data Frame之间的相关系数 # axis=1进行行进行计算 # print obj.corrwith(obj.IBM) # 唯一值 obj1 = Series(["c",'a','d','a','a','b','b','c','c']) uniques = obj1.unique() # 加排序 # print uniques.sort() # 计算出现的频率 print obj1.value_counts()