Please double check that the object lg corresponds to LogisticRegression
#Help with an error
dreamy merlin
muted night
i checked it
fiery hound
Look, this is not a regression task it's a classification task
regression would be for predicting an item in a sequence or continuous function. Detecting emotion is more along the lines of classification
You even have it in your code as "label"
muted night
so what should I do ?
fiery hound
what is the model you are training to classify?
muted night
Logistic Regression
fiery hound
change it to a classifier
what is the model you import?
Give the import line + model initialization code
muted night
fiery hound
Ok.... that's not helpful
muted night
fiery hound
still not helpful
you loaded the lg variable from a pkl file. That doesn't tell me what the model is
muted night
fiery hound
the model could be a SVM, KNN, random tree/forest, MLP/NN, etc
muted night
you loaded the `lg` variable from a pkl file. That doesn't tell me what the mode...
I'll share the jupyter file
fiery hound
Looking at your error, go back and make sure your df_test is the same as your df_train after you transform both with the TF-IDF vectorizer
muted night
oh okay
fiery hound
the error you gave points to the vectorizor calling predict() which is not right
muted night
yep but it says tfidfvectorizer has no attribute predict
fiery hound
yes, that's correct
muted night
but predict is being used with logistic regression model
muted night
yup
fiery hound
but for some reason, the program is seeing lg as a TFIDFVectorizer -> which does NOT have .predict()
muted night
yeah
fiery hound
What happens when you output type(tf_test) and type(tf_train)?
I ask because you use the vectorizor on those datasets differently.
you fit_transform on train data but only transform on the test data
muted night
What happens when you output `type(tf_test)` and `type(tf_train)`?
scipy.sparse._csr.csr_matrix
fiery hound
for both?
muted night
yup
fiery hound
what about type(lg)?
muted night
what about `type(lg)`?
sklearn.linear_model._logistic.LogisticRegression
fiery hound
Hmmm.....
@muted night do the same for inpvector
It's odd that the part that's throwing an error is a function you never call in your code.
muted night
<@1170788165263560744> do the same for `inpvector`
list
muted night
It's odd that the part that's throwing an error is a function you never call in ...
its inside a function i executed the function on string seperately
muted night
when the other data is a csr matrix?
they are trained on matrix that contains sentences and emotion
fiery hound
Right but after your tfidf vectorizer transforms the text, it should be in csr/csc matrix
fiery hound
Yeah but you have to print the type of inpvector from within the function
otherwise inpvector is going to be None because the variable is out of scope (of the function)
muted night
<class 'sklearn.linear_model._logistic.LogisticRegression'>
fiery hound
Hmmm... Rerun it all again
muted night
okayy
fiery hound
because in your screenshot, you should have gotten the error again when you called your function but it ran just fine.
muted night
done
fiery hound
Any error messages?
muted night
nope
fiery hound
Then I think you're all set