Please run the code to see entire visualizations and comments of each tasks!

Dataset is a subset of ABSITA dataset.

• The data includes reviews and opinion (negative/positive) about hotels listed on Booking.com website in 3 cities: Napoli, Bologna and Milan.
• The reviews are in multi-languages but we're going to analyze exclusively Italian reviews.
• We consider text reviews of customers and labeled sentiments encoded to the 'target' column.

• Preprocessing the data by splitting sentences, splitting unigram or n-grams, removing stop words and punctuations, and forming a term-document matrix using tf-idf indexes.
• Build a pipeline for preprocessing steps.
• Classification: Applied Logistic Regression and Random Forest classification methods and executed a GridSearchCV cross Validation in order to find the best parameters and the best score. The labels are negative or positive.
• Clustering: Using K-means and Hierarchical Agglomerative Cluster methods.
• Topic Modeling: Extracing top 3 and 8 topics of the reviews, listin representiative of each topics and visualizing them using pyLDAvis library.
• Summarization: summary reviews to a consise and meaningful paragrahp of 50 words.

Quynh-Anh Ng