使用scikit-learn进行聚类

Question

I'm working with scikit-learn for the first time and am trying to do a kmeans cluster. I think I'm doing it all correctly.

I have a datetime index and 2 columns of ints in a dataframe df .

kmeans = KMeans(n_clusters=2)
kmeans.fit(df.values)

Then I have another set of data that looks the same and I want to predict it. So I pass df1 into kmeans.predict() .

Do I need to add some column to each of those dataframes for the classification? I'm assuming everything I put into the fit is good.

After getting a classification completed, how do I then visualize it in a graph?

Thanks

Answer 1

Without seeing the data and assuming you want the resultant prediction as a column in the second data (df2) frame you can apply the kn.predict() using the .apply() function and specifying the vertical axis. This will give you an additional column with the predicted output.

ie

df2['predictions'] = df2['values'].apply(kmeans.predict)

Heres the info on apply. http://pandas.pydata.org/pandas-docs/version/0.17.1/generated/pandas.DataFrame.apply.html

Hope that helps. Let me know if you need anything else.