Logprior is driving me crazy

Course Q&A Natural Language Processing NLP with Classification and Vector Spaces
,
1

I don’t understand the error below. Why is the verification function expecting something else than 0 ?

image
image721×403 12.7 KB

Thank you for your help

2

Here’s another recent thread on this assignment. Please have a look and see if that sheds any light on your issues. In one of my replies there, I show some debugging output that is worth a look.

1 Like
3

@paulinpaloalto could you explain why this error is coming ? I am not able to resolve this

TIA

4

There must be something wrong in your implementation. It’s a bit hard to diagnose with the input “I have a problem”. :nerd_face: Please note that the mentors do not have the superpower to examine anyone else’s notebooks. What is your output equivalent to what I show on the other thread I referenced above?

5

image

These are my outputs

6 
V = 9165, len(wordlist) 11436
V: 9165, V_pos: 5804, V_neg: 5632, D: 8000, D_pos: 4000, D_neg: 4000, N_pos: 27547, N_neg: 27152
freq_pos for smile = 47
freq_neg for smile = 9
loglikelihood for smile = 1.5577981920239676
0.0
9165

That’s good: your numbers look correct for N_pos and N_neg. How about the specific numbers for “smile”?

7

image
image711×115 11.8 KB

They all are matching yet the test function does not passes all cases

8

Hmmm, that is correct as far as I can see. Time to use the “in case of emergency, break glass” method. Please check your Direct Messages.

9

It turns out your error message (which you never actually showed us) is exactly the same as the one in the original post on this thread. Note that the logprior value is only based on D_pos and D_neg. In your case and the OP, your answer for logprior is 0.0, meaning that D_pos and D_neg are equal. In the test case in the notebook, that is actually true, but in the test imported from the public test file, that is no longer true. So there must be a bug in your logic that causes D_pos and D_neg to be equal when they shouldn’t be. So that is a pretty big clue as to where to look for the error.

Here is the complete output I get with my extra debugging statements for the test case from the imported unit tests:

V: 9165, V_pos: 5804, V_neg: 5632, D: 8000, D_pos: 4000, D_neg: 4000, N_pos: 27547, N_neg: 27152
freq_pos for smile = 47
freq_neg for smile = 9
loglikelihood for smile = 1.5577981920239676
V: 9165, V_pos: 5804, V_neg: 5632, D: 20, D_pos: 10, D_neg: 10, N_pos: 27547, N_neg: 27152
freq_pos for smile = 47
freq_neg for smile = 9
loglikelihood for smile = 1.5577981920239676
V: 9165, V_pos: 5804, V_neg: 5632, D: 15, D_pos: 10, D_neg: 5, N_pos: 27547, N_neg: 27152
freq_pos for smile = 47
freq_neg for smile = 9
loglikelihood for smile = 1.5577981920239676
 All tests passed

So you can see that there are a couple of test cases where D_pos and D_neg are not equal.