Question

我正在使用一个名为pysal的包来运行以下Theil Decomposition，以在输出之内和之间找到它们。

当我在包装下面创建一个小的数据框时，它起作用。参见下面的代码：

import pysal 
path="/Users/username/Desktop/file1.csv"
df_table=pd.read_table(path, sep=",")  
df2=pd.DataFrame(df_table)
df= df2.sort_values(['exposure'], ascending=True)

rr = np.array(df['exposure'])
drop = pysal.inequality.theil.Theil(rr)
print ('drop.T', drop.T) # this is total theil 

dropp = pysal.inequality.theil.TheilD(rr, df['race'] )
print ('WG', dropp.wg) #within group
print ("BG", dropp.bg) #between group

当我尝试在更大的文件上运行相同的代码时，出现以下错误消息：

如何修复错误消息？

TypeError: unorderable types: float() < str()

下面是pysal包的源代码两种文件类型的数据类型似乎相同。我不确定为什么它会工作一个小文件，而不是一个大文件。

def __init__(self, y, partition):
    groups = np.unique(partition)
    T = Theil(y).T
    ytot = y.sum(axis=0)

    #group totals
    gtot = np.array([y[partition == gid].sum(axis=0) for gid in groups])
    mm = np.dot

    if ytot.size == 1:  # y is 1-d
        sg = gtot / (ytot * 1.)
        sg.shape = (sg.size, 1)
    else:
        sg = mm(gtot, np.diag(1. / ytot))
    ng = np.array([sum(partition == gid) for gid in groups])
    ng.shape = (ng.size,)  # ensure ng is 1-d
    n = y.shape[0]
    # between group inequality
    sg = sg + (sg==0) # handle case when a partition has 0 for sum
    bg = np.multiply(sg, np.log(mm(np.diag(n * 1. / ng), sg))).sum(axis=0)
    self.T = T
    self.bg = bg
    self.wg = T - bg

大型数据集的Python3 pysal错误消息

0 个答案: