BUG: to_latex with multicolumn and multiindex joins cells which are on different hierarchy levels

[wlter]

Code Sample, a copy-pastable example if possible

import pandas as pd
import numpy as np


df = pd.DataFrame()


idx = 0
for i in range(2):
  for val0 in [3,2,1]:
    for val1 in range(val0):
      idx = idx + 1
      r0 = str(np.mod(int(idx*0.5),2))
      r1 = np.random.uniform()
      df = df.append({"i":i,"val0" : val0,\
                            "val1":val1, 
                            "r0":r0, \
                          "r1":r1}, ignore_index=True)


df.set_index(["i",'val0','val1',"r0", 'r1'], inplace=True)

print(df.to_latex(multirow = True, escape=False))

produces

\begin{tabular}{lllll}
	\toprule
	&     &     &   &          \\
	i & val0 & val1 & r0 & r1 \\
	\midrule
	\multirow{6}{*}{0.0} & \multirow{3}{*}{3.0} & 0.0 & 0 & 0.307919 \\
	&     & 1.0 & \multirow{2}{*}{1} & 0.488816 \\
	&     & 2.0 &   & 0.708405 \\
	\cline{2-5}
	\cline{4-5}
	& \multirow{2}{*}{2.0} & 0.0 & \multirow{2}{*}{0} & 0.806916 \\
	&     & 1.0 &   & 0.763446 \\
	\cline{2-5}
	\cline{4-5}
	& 1.0 & \multirow{2}{*}{0.0} & \multirow{2}{*}{1} & 0.255642 \\
	\cline{1-5}
	\multirow{6}{*}{1.0} & \multirow{3}{*}{3.0} &     &   & 0.093269 \\
	\cline{3-5}
	\cline{4-5}
	&     & 1.0 & \multirow{2}{*}{0} & 0.775120 \\
	&     & 2.0 &   & 0.989241 \\
	\cline{2-5}
	\cline{4-5}
	& \multirow{2}{*}{2.0} & 0.0 & \multirow{2}{*}{1} & 0.741230 \\
	&     & 1.0 &   & 0.960813 \\
	\cline{2-5}
	\cline{4-5}
	& 1.0 & 0.0 & 0 & 0.559090 \\
	\bottomrule
\end{tabular}

produces

Problem description

Hey,

when using a multi-index dataframe, the to_latex command with option "multirow = True" joins cells, which due to the multirow hierarchy shouldn't be joined. In the image this produces the joined entry on the divider line.

Expected Output

• cells should not be joint if the hierarchy divides them

Output of pd.show_versions()

INSTALLED VERSIONS ------------------ commit: None python: 3.5.3.final.0 python-bits: 32 OS: Windows OS-release: 7 machine: AMD64 processor: AMD64 Family 16 Model 4 Stepping 3, AuthenticAMD byteorder: little LC_ALL: None LANG: en LOCALE: None.None

pandas: 0.20.2
pytest: 3.0.7
pip: 9.0.1
setuptools: 27.2.0
Cython: 0.25.2
numpy: 1.12.1
scipy: 0.19.0
xarray: None
IPython: 5.3.0
sphinx: 1.5.6
patsy: 0.4.1
dateutil: 2.6.0
pytz: 2017.2
blosc: None
bottleneck: 1.2.1
tables: 3.2.2
numexpr: 2.6.2
feather: None
matplotlib: 2.0.2
openpyxl: 2.4.7
xlrd: 1.0.0
xlwt: 1.2.0
xlsxwriter: 0.9.6
lxml: 3.7.3
bs4: 4.6.0
html5lib: 0.999
sqlalchemy: 1.1.9
pymysql: None
psycopg2: None
jinja2: 2.9.6
s3fs: None
pandas_gbq: None
pandas_datareader: None

[wlter]

So, I wrote my own script, which is in no way compatible - but maybe the idea helps here: I used a hierarchical multi-index for the first levels and left the rest of the hierarchy to default dataframes. In the table, the hierrachichal elements are all distributed over mutliple cells using multirow. The hierarchical elemnts are also creating the clines. I don't allow multirow in the area of default dara frames. So my suggestion would be to not join cells because of their content but:

on the level of the multiindex:

• join cells when they belong to the same hierarchical element using multirow
• add the necessary \clines

on the level of the dataframes (below the multiindex)

• do not join cells

TLDR: Suggesting to join cells based on whether they are part of multi-index or default dataframe instead of their content

[sgsaenger]

Does this problem persist in any way?
v0.23.4 seems to correctly yield:

\begin{tabular}{lllll}
\toprule
    &     &     &   &          \\
i & val0 & val1 & r0 & r1 \\
\midrule
\multirow{6}{*}{0.0} & \multirow{3}{*}{3.0} & 0.0 & 0 & 0.604922 \\
    &     & 1.0 & 1 & 0.204031 \\
    &     & 2.0 & 1 & 0.140646 \\
\cline{2-5}
    & \multirow{2}{*}{2.0} & 0.0 & 0 & 0.456303 \\
    &     & 1.0 & 0 & 0.799159 \\
\cline{2-5}
    & 1.0 & 0.0 & 1 & 0.328993 \\
\cline{1-5}
\multirow{6}{*}{1.0} & \multirow{3}{*}{3.0} & 0.0 & 1 & 0.078872 \\
    &     & 1.0 & 0 & 0.489851 \\
    &     & 2.0 & 0 & 0.058885 \\
\cline{2-5}
    & \multirow{2}{*}{2.0} & 0.0 & 1 & 0.889743 \\
    &     & 1.0 & 1 & 0.498069 \\
\cline{2-5}
    & 1.0 & 0.0 & 0 & 0.397285 \\
\bottomrule
\end{tabular}

[mroeschke]

Looks to produce a correct output on master. Could use a test:

In [7]: df = pd.DataFrame()
   ...:
   ...:
   ...: idx = 0
   ...: for i in range(2):
   ...:   for val0 in [3,2,1]:
   ...:     for val1 in range(val0):
   ...:       idx = idx + 1
   ...:       r0 = str(np.mod(int(idx*0.5),2))
   ...:       r1 = np.random.uniform()
   ...:       df = df.append({"i":i,"val0" : val0,\
   ...:                             "val1":val1,
   ...:                             "r0":r0, \
   ...:                           "r1":r1}, ignore_index=True)
   ...:
   ...:
   ...: df.set_index(["i",'val0','val1',"r0", 'r1'], inplace=True)
   ...:
   ...: print(df.to_latex(multirow = True, escape=False))
\begin{tabular}{lllll}
\toprule
    &     &     &   &          \\
i & val0 & val1 & r0 & r1 \\
\midrule
\multirow{6}{*}{0.0} & \multirow{3}{*}{3.0} & 0.0 & 0 & 0.959256 \\
    &     & 1.0 & 1 & 0.617433 \\
    &     & 2.0 & 1 & 0.355065 \\
\cline{2-5}
    & \multirow{2}{*}{2.0} & 0.0 & 0 & 0.493983 \\
    &     & 1.0 & 0 & 0.710831 \\
\cline{2-5}
    & 1.0 & 0.0 & 1 & 0.068669 \\
\cline{1-5}
\multirow{6}{*}{1.0} & \multirow{3}{*}{3.0} & 0.0 & 1 & 0.391476 \\
    &     & 1.0 & 0 & 0.630937 \\
    &     & 2.0 & 0 & 0.892755 \\
\cline{2-5}
    & \multirow{2}{*}{2.0} & 0.0 & 1 & 0.018855 \\
    &     & 1.0 & 1 & 0.444306 \\
\cline{2-5}
    & 1.0 & 0.0 & 0 & 0.590894 \\
\bottomrule
\end{tabular}


In [8]: pd.__version__
Out[8]: '0.26.0.dev0+627.gef77b5700'

[mroeschke]

Actually this looks incorrect again:

In [1]:         df = pd.DataFrame(
   ...:             index=[(0.0, 3.0, 0.0, 0, 0.929121402530996), (0.0, 3.0, 1.0, 1, 0.4187460393711979),
   ...:                    (0.0, 3.0, 2.0, 1, 0.47309,
   ...:                     69460209665), (0.0, 2.0, 0.0, 0, 0.3044652225366862),
   ...:                    (0.0, 2.0, 1.0, 0, 0.8087530860532746), (0.0, 1.0, 0.0, 1, 0.0799379522,
   ...:                                                             5657385),
   ...:                    (1.0, 3.0, 0.0, 1, 0.7793463916039404), (1.0, 3.0, 1.0, 0, 0.21066478200369132),
   ...:                    (1.0, 3.0, 2.0, 0, 0.3143737387268,
   ...:                     193), (1.0, 2.0, 0.0, 1, 0.46081170223017887),
   ...:                    (1.0, 2.0, 1.0, 1, 0.8865341655631166), (1.0, 1.0, 0.0, 0, 0.31459248512345084
   ...:                                                             )])
   ...:         result = df.to_latex(multirow = True, escape=False)

# should be `begin{tabular}{lllll}`
In [2]: result
Out[2]: "\\begin{tabular}{l}\n\\toprule\nEmpty DataFrame\nColumns: Index([], dtype='object')\nIndex: Index([    (0.0, 3.0, 0.0, 0, 0.929121402530996),\n          (0.0, 3.0, 1.0, 1, 0.4187460393711979),\n        (0.0, 3.0, 2.0, 1, 0.47309, 69460209665),\n          (0.0, 2.0, 0.0, 0, 0.3044652225366862),\n          (0.0, 2.0, 1.0, 0, 0.8087530860532746),\n       (0.0, 1.0, 0.0, 1, 0.0799379522, 5657385),\n          (1.0, 3.0, 0.0, 1, 0.7793463916039404),\n         (1.0, 3.0, 1.0, 0, 0.21066478200369132),\n        (1.0, 3.0, 2.0, 0, 0.3143737387268, 193),\n         (1.0, 2.0, 0.0, 1, 0.46081170223017887),\n          (1.0, 2.0, 1.0, 1, 0.8865341655631166),\n         (1.0, 1.0, 0.0, 0, 0.31459248512345084)],\n      dtype='object') \\\\\n\\bottomrule\n\\end{tabular}\n"

In [3]: pd.__version__
Out[3]: '1.0.0rc0+134.gbbcda98c7.dirty'