Naive Bayes Gaussian Classification Prediction not taking array in SK-Learn

Question

I have made the following gaussian prediction model in SK-learn:

chess_gnb = GaussianNB().fit(raw[['elo', 'opponent_rating', 'winner_loser_elo_diff']],raw['winner'])

I then made a test array and attempted to feed it into the model:

test1 = [['elo', 1000], ['opponent_rating', 800], ['winner_loser_elo_diff', 200]]
chess_gnb.predict(test1)

However, I'm getting this error:

ValueError: Unable to convert array of bytes/strings into decimal numbers with dtype='numeric'

The 'winner' prediction should be a string that can have one of two values. Why am I getting the valueError if all of my inputs are integers?

Answer 1

You need to provide a dataframe, using an example:

import pandas as pd
import numpy as np
from sklearn.naive_bayes import GaussianNB

np.random.seed(123)

raw = pd.DataFrame(np.random.uniform(0,1000,(100,3)),
                   columns = ['elo','opponent_rating','winner_loser_elo_diff'])
raw['winner'] = np.random.binomial(1,0.5,100)

chess_gnb = GaussianNB().fit(raw[['elo', 'opponent_rating', 'winner_loser_elo_diff']],raw['winner'])

This works:

test1 = pd.DataFrame({'elo': [1000],'opponent_rating':[800],'winner_loser_elo_diff':[200]})
chess_gnb.predict(test1)