sklearn.neighbors.KNeighborsClassifier could not convert string to float

Question

I am trying to clean my data in python using sklearn.neighbors.KNeighborsClassifier. In the fit function of classifier I have provide training data in the form of a DataTable generated by Pandas from a csv file.

The fit function throws an error

could not convert string to float : "training data's first cell value".

How can I use the training data of tabular form of strings?

I am working on Kaggle Titanic dataset.

No the columns are strings and i am interested to train my classifier with the string data. — Zeshan Khan, Oct 25 '17 at 09:45
You can't train it directly with strings. You need to transform the text data into numerical values. Algorithms can only understand numbers. Show us sample data and what you have done so far to help better. — yesemsanthoshkumar, Oct 25 '17 at 10:50

score 1 · Accepted Answer · answered Oct 25 '17 at 09:50

1

You have to define a metrics for your classifier. Otherwise, the clustering does not have a clue, what he has to do with strings. Look at this thread, which is probably the same:

https://stackoverflow.com/a/35283104/2151532

answered Oct 25 '17 at 09:50

loose11

607
6
31

sklearn.neighbors.KNeighborsClassifier could not convert string to float

1 Answers1