Pandas warnings when formating dataframe as event log

[fmannhardt]

I am trying to import a CSV file (sepsis.csv) using the following approach:

sepsis = pd.read_csv("sepsis.csv", sep=';')
sepsis_log = pm4py.format_dataframe(sepsis, case_id='case_id', activity_key='activity', timestamp_key='timestamp')
sepsis_log = pm4py.convert_to_event_log(sepsis_log)

The following warnings show up:

XXX\pm4py\utils.py:87: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  df[constants.CASE_CONCEPT_NAME] = df[constants.CASE_CONCEPT_NAME].astype("string")

XXXX\pm4py\utils.py:89: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  df[xes_constants.DEFAULT_NAME_KEY] = df[xes_constants.DEFAULT_NAME_KEY].astype("string")

The sepsis file is attached but nothing special in there. I am using Python 3.8 and the packages versions installed are:

Package            Version
------------------ ----------
asttokens          2.1.0     
backcall           0.2.0     
backports.zoneinfo 0.2.1     
certifi            2022.6.15 
colorama           0.4.6     
contourpy          1.0.6     
cvxopt             1.3.0     
cycler             0.11.0    
debugpy            1.6.3     
decorator          5.1.1     
deprecation        2.1.0     
entrypoints        0.4       
executing          1.2.0     
fonttools          4.38.0    
graphviz           0.20.1    
intervaltree       3.1.0     
ipykernel          6.17.0    
ipython            8.6.0     
ipywidgets         8.0.2     
jedi               0.18.1    
Jinja2             3.1.2
joblib             1.2.0
jsonpickle         2.2.0
jupyter_client     7.4.4
jupyter_core       4.11.2
jupyterlab-widgets 3.0.3
kiwisolver         1.4.4
lxml               4.9.1
MarkupSafe         2.1.1
matplotlib         3.6.1
matplotlib-inline  0.1.6
mizani             0.8.1
mpmath             1.2.1
nest-asyncio       1.5.6
networkx           2.8.7
numpy              1.23.4
packaging          21.3
palettable         3.3.0
pandas             1.5.1
parso              0.8.3
patsy              0.5.3
pickleshare        0.7.5
Pillow             9.3.0
pip                22.2.2
plotnine           0.10.1
pm4py              2.2.30
prompt-toolkit     3.0.31
psutil             5.9.3
pure-eval          0.2.2
pydotplus          2.0.2
Pygments           2.13.0
pyparsing          3.0.9
python-dateutil    2.8.2
pytz               2022.6
pyvis              0.3.0
pywin32            304
pyzmq              24.0.1
scikit-learn       1.1.3
scipy              1.9.3
setuptools         65.5.0
six                1.16.0
sklearn            0.0
sortedcontainers   2.4.0
stack-data         0.6.0
statsmodels        0.13.2
StringDist         1.0.9
sympy              1.11.1
threadpoolctl      3.1.0
torch              1.13.0+cpu
tornado            6.2
tqdm               4.64.1
traitlets          5.5.0
typing_extensions  4.4.0
tzdata             2022.6
wcwidth            0.2.5
wheel              0.37.1
widgetsnbextension 4.0.3
wincertstore       0.2

[fit-alessandro-berti]

Dear Felix,

Yes I am aware of that warning. The only thing the row do is to replace the content of the column with itself typed as string. Actually I am not aware of a better method within Pandas.

In release 2.2.30 we deprecated the method format_dataframe though, in favor of a bigger freedom to the user, that needs to format manually the timestamp columns. Also the methods are always accepting the case, activity and timestamp column (so no need to coerce them into case:concept:name, concept:name and time:timestamp )

[fmannhardt]

Do you have a link on the best practice on how to use Pandas data frames from version 2.2.30?

[fit-alessandro-berti]

Sorry my mistake. I meant the upcoming version 2.3.0 (of which we released the rc2)

First we will complete the release process of 2.3.0 and then update the documentation. For some time, documentation in the web site will be overlapping (so every method reported should work for both 2.2.x and 2.3.x). Then it will be more specialized. For the first times, it will be still suggested to use the format_dataframe method even if deprecated.

[fmannhardt]

Ok, then I keep using the method since the course starts in a week. Not a Pandas expert but I found this information: https://stackoverflow.com/questions/20625582/how-to-deal-with-settingwithcopywarning-in-pandas So, maybe the method could avoid the warning by locally using pd.options.mode.chained_assignment = None :-)