Question

我有一个CSV文件，它将每个国家/地区映射到某个值，但问题是它没有很好地形成，它的标题有重复的模式：国家/地区，金额，国家/地区，金额，...... 。（此处金额衡量的是不同的东西，例如自杀率，酒精消费量等，请注意，对于某些国家/地区的数据缺失），请参阅输入DataFrame：df_in。

我希望将国家/地区作为索引以及那些＆＃39; Amounts＆＃39;作为列，请参阅输出DataFrame，df_out

df_in = pd.read_csv('https://dl.dropboxusercontent.com/u/40513206/input.csv', sep = ';', header = 0, index_col = None,
             na_values = [''], mangle_dupe_cols = False)

df_out = pd.read_csv('https://dl.dropboxusercontent.com/u/40513206/output.csv', sep = ';', header = 0, index_col = None,
             na_values = [''], mangle_dupe_cols = False)

我原本以为我首先从输入中获取所有唯一的国家/地区（例如，将其作为新的空数据框架的索引）

col_pat = df_in.columns[df_in.columns.to_series().str.contains('Countries')]
cntry = df_in.ix[:, col_pat]
un_elm = pd.Series(map(str, pd.unique(cntry.values.ravel())))
countries = un_elm[un_elm != 'nan']

然后开始拆分主DataFrame（Counrtries as index和Amount as column）并将其累加到空DataFrame。还有其他想法，谢谢？

Answer 1

首先使用.ix根据位置选择列

df_in = pd.read_csv('https://dl.dropboxusercontent.com/u/40513206/input.csv', sep = ';',      header = 0, index_col = None,
         na_values = [''], mangle_dupe_cols = False)

df1 = df_in.ix[:,:2].dropna().set_index('Countries1')
df2 = df_in.ix[:,2:4].dropna().set_index('Countries2')
df3 = df_in.ix[:,4:].dropna().set_index('Countries3')

然后在轴1上连接：

pd.concat([df1,df2,df3], axis=1)


               Amount  Amount  Amount
Austria           NaN       5     NaN
Denmark             6     NaN     NaN
France              3     NaN     NaN
Ireland           NaN     NaN       6
Norway            NaN       2     NaN
Russia            NaN     NaN       5
Slovenia          NaN     NaN       4
Spain             NaN       3       3
Sweden              5       1       2
Switzerland         4       4     NaN
U.K.                1     NaN     NaN
United States       2     NaN       1

解析复杂的csv文件

1 个答案: