python - 将所有索引列合并并排序到单个数据帧的单个列中

Question

我有一个看起来像这样的数据框。它有更多时间轴，直到 Time[s].30。

Time[s]    v1   Time[s].1   v2
160.84621   0   160.84808   7
161.14613   0   161.14802   7
161.538245  27  161.540085  7
162.01598   27  162.017865  7
162.31589   27  162.317775  7
162.615855  27  162.617735  7
162.915765  27  162.91765   7
163.21574   27  163.217625  7
163.51569   27  163.517575  7
163.81563   27  163.81751   7
164.11554   27  164.117425  7
164.4155    27  164.41738   9
164.71543   27  164.717315  9
165.015405  27  165.017285  9
165.31532   27  165.317205  9
165.65083   26  165.65272   9
165.95025   26  165.95214   9

我想要一个时间轴Time[s].general，它是所有带有排序值的时间列的合并形式。我已对所有这些列建立了索引。

df.set_index(keys=list(file_read.filter(like='Time[s]').columns))

更新:

预期输出:

Time[s]      v1     v2
160.84621   0      null 
160.84808   null     7
160.14613   0      null
161.14802   null     7
161.538245  27     null
161.540085  null     7
162.01598   27     null
162.017865  null     7
162.31589   27     null
162.317775  null     7

等等。

更新2:

Time[s]    v1   Time[s].1   v2      Time[s].2   v3
160.84621   0   160.84808   7   158.538395  Active
161.14613   0   161.14802   7   158.538515  Active
161.538245  27  161.540085  7   159.49455   Active
162.01598   27  162.017865  7   162.352395  Locked
162.31589   27  162.317775  7   163.35075   Locked
162.615855  27  162.617735  7   164.350675  Locked
162.915765  27  162.91765   7   165.350655  Locked
163.21574   27  163.217625  7   166.509695  Locked
163.51569   27  163.517575  7   166.509815  Locked
163.81563   27  163.81751   7   167.50086   Locked
164.11554   27  164.117425  7   168.50085   Locked
164.4155    27  164.41738   9   169.500865  Locked
164.71543   27  164.717315  9   171.502655  Standby
165.015405  27  165.017285  9   185.89923   Forward
165.31532   27  165.317205  9   3273.448065 Forward
165.65083   26  165.65272   9   3274.43487  Forward
165.95025   26  165.95214   9   3275.4348   Forward

Answer 1

我认为需要:

b  = df.filter(like='v').columns

d = {x: 'v.{}'.format(i) for i, x in enumerate(b)}
d['Time[s]'] = 'Time[s].0'
print (d)
{'v1': 'v0', 'v2': 'v1', 'Time[s]': 'Time[s].0'}

df = df.rename(columns=d)
L = [x.set_index(x.columns[0]) for i, x in df.groupby(lambda x: x.split('.')[-1], axis=1)]
df = pd.concat(L, axis=1)
print (df.head(10))
             v.0  v.1
160.846210   0.0  NaN
160.848080   NaN  7.0
161.146130   0.0  NaN
161.148020   NaN  7.0
161.538245  27.0  NaN
161.540085   NaN  7.0
162.015980  27.0  NaN
162.017865   NaN  7.0
162.315890  27.0  NaN
162.317775   NaN  7.0

扩展:

1. 第一filter columns 字典的所有 v 列，用于时间戳与值列对。
2. rename by dict，也是第一个timestamp列
3. groupby 按 . 后的 columna 值进行列表理解，按 set_index 创建索引和 concat一起

编辑:

如果数值和重复时间戳聚合是按 mean 聚合，如果不是，则按 first 聚合:

b  = df.filter(like='v').columns

d = {x: 'v.{}'.format(i) for i, x in enumerate(b)}
d['Time[s]'] = 'Time[s].0'
print (d)
{'v1': 'v0', 'v2': 'v1', 'Time[s]': 'Time[s].0'}

df = df.rename(columns=d)
L = [x.groupby(x.columns[0]).mean() 
     if np.issubdtype(df[x.columns[1]].dtype, np.number)
     else x.groupby(x.columns[0]).first() 
     for i, x in df.groupby(df.columns.str.split('.').str[-1], axis=1)]

df = pd.concat(L, axis=1)
print (df.head(10))
             v.0  v.1     v.2
158.538395   NaN  NaN  Active
158.538515   NaN  NaN  Active
159.494550   NaN  NaN  Active
160.846210   0.0  NaN     NaN
160.848080   NaN  7.0     NaN
161.146130   0.0  NaN     NaN
161.148020   NaN  7.0     NaN
161.538245  27.0  NaN     NaN
161.540085   NaN  7.0     NaN
162.015980  27.0  NaN     NaN