Properly handle columns with mixed numeric and categorical values about ydata-profiling

[romainx]

Please could you provide an example permitting to reproduce the case?

Note: Since #43, it is possible to call the report with the option check_correlation=False to skip this phase. In your case, it will avoid the error since it appears during the correlation computing.

[navado]

Example data: ex.txt
More observations - if I reduce the size of the example to ~1000 points, including the problematic records, the problem does not happen.
Another strange thing, enabling check_correlation=False throws an error

>>> pfr = pandas_profiling.ProfileReport(subset,check_correlation=False)
.....
TypeError: _plot_histogram() got an unexpected keyword argument 'check_correlation'

[conradoqg]

Hey,

This issue was fixed in the PR #82.

I think we can close this issue after those changes.

Best