0
$\begingroup$

I am running Python code off Kaggle on the adult dataset using Naive Bayes. How do you interpret the results below, I know that it's for each instance the likelihood they make less than 50K or more than 50k.

How do you interpret say the first row in simple terms? Is there a way I could make it in standard form?

y_pred_prob_df = pd.DataFrame(data=y_pred_prob, columns=['Prob of - <=50K', 'Prob of - >50K'])

y_pred_prob_df

enter image description here

Improve this question
$\endgroup$
4
  • $\begingroup$ What do you mean with 'standard form'? The first row simply means that for the first observation the model predicts a probabiliy of 99.99% (basically 100%) that they make less than 50k. $\endgroup$
    – Oxbowerce
    Commented Feb 9, 2022 at 10:39
  • $\begingroup$ Thanks. So row 2 [1], predicts a probably of 99% that they make less than 50k AND 31% that will make more than 50k? @Oxbowerce $\endgroup$
    – FredNina
    Commented Feb 9, 2022 at 10:55
  • $\begingroup$ It indeed predicts a probability of 99.97% to make less than 50k, but the probability of making more than 50k is not 31% but 0.0312% (notice the e-4). The two numbers always add up to 100%. $\endgroup$
    – Oxbowerce
    Commented Feb 9, 2022 at 11:12
  • $\begingroup$ @Oxbowerce Thank you! that makes way more sense now . Much appreciated . $\endgroup$
    – FredNina
    Commented Feb 9, 2022 at 11:26

1 Answer 1

Reset to default
0
$\begingroup$

The first row simply means that for the first observation the model predicts a probabiliy of 99.99% (basically 100%) that they make less than 50k. For the second records your model predicts a probability of 99.97% to make less than 50k, but the probability of making more than 50k is not 31% but 0.0312% (notice the e-4). The two numbers always add up to 100%.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.