Search code examples
pythonnumpyscipyscikit-learnsklearn-pandas

unorderable types error when importing sklearn

I installed numpy(1.12.0b1), Scipy(0.18) on windows. I also installed sci-kit as well. When i wrote "import sklearn" in python console, it gives an error like this: if np_version < (1, 12, 0): TypeError: unorderable types: str() < int() What will be the issue?

Solution

  • The problem is out on the version number, so maybe you could try to revise fixs.py in the sklearn folder. Add these script after the try in line 32:

    if not (x.isdigit()):
    x='0'

    so your codes will be:

    def _parse_version(version_string):
version = []
for x in version_string.split('.'):
    try:
        if not (x.isdigit()):
            x='0'
        version.append(int(x))
        #print(x)
    except ValueError:
        # x may be of the form dev-1ea1592
        version.append(x)
return tuple(version)