Explanation

IMDB movie reviews

This project focuses on Imdb (Internet Mobile DataBase) movie reviews using NLP.
Dataset loading:
Loaded "imdb" dataset from huggingface, it containing movie reviews text with sentiment labels 0=Negative,1=Positive.
Text Preprocessing:
Extract review texts and used nltk for cleaning. Nltk source to download it stopwords,wordnet,omw for lemmatization. Cleaned texts apply in new column.
Data split:
split dataset into train ,validation sets.
Features extraction:
Clean words tfidf to convert into numerical features.
Model trainning: using classification algorithms logistic regression,naive bayes(multinomalNB) train a models.
Evaluation:
accuracy,test accuracy for the model performance and comparision. Also classification report precision FP,recall FN,f1 score.
Final use:
Trained model can new_user reviews input to classify as Positive or Negative

Result

Logistic regression achieve the validation accuracy 88% and test accuracy 87%. Logistic regression classifier high precision,recall ,f1 score fewer of FP,FN then perform well.

streamlit web app for clear visual:

Positive Review:

Negative Review: