Sklearn.PCA-unexpected keyword argument 'random_state'

Question

def doPCA(data, dimensions=2):
from sklearn.decomposition import PCA
model = PCA(n_components=dimensions, svd_solver='randomized', random_state=7)
model.fit(data)
return model

File "/home/dogus/Downloads/DAT210x-master/Module5/assignment4.py", line 221, in display_pca = doPCA(T)

File "/home/dogus/Downloads/DAT210x-master/Module5/assignment4.py", line 56, in doPCA model = PCA(n_components=dimensions, svd_solver='randomized', random_state=None)

TypeError: init() got an unexpected keyword argument 'random_state'

• I use Spyder with python2.7 in Ubuntu.
• I installed sklearn with conda,i also installed it via 'pip install'.
• I tried random_state=None but gave me the same error.
• Then i ran it via terminal and it worked without any error.

Edit:

• Scikit-learn(1.18.1),spyder,scipy,numpy,ipython updated via anaconda.
• Then i restarted Spyder and tried to run the code but gave me the same error.
• I ran the codes via IPython and Python consoles.

Problem caused by Spyder? How can i fix it?

Answer 1

You are probably using an old version of sklearn. The docs say that the random_state variable was added in version 0.18.0. Try updating sklearn.

In order to make the update, look at these questions

• How to upgrade scikit-learn package in anaconda

• conda update scikit-learn (also scipy and numpy)

---

Based on your comment, just to make sure the problem is not the scikit-learn version, try the following

def doPCA(data, dimensions=2): 
    from sklearn.decomposition import PCA
    import sklearn
    print sklearn.__version__
    model = PCA(n_components=dimensions, svd_solver='randomized', random_state=7)
    model.fit(data)
    return model

This will only print the version of sklearn being used. As you said that you had also installed sklearn with pip, you could have conflicting versions.