Natural Language Processing AND Sentiments Analysis of Disney tweets
Project Overview and objectives:
The Walt Disney Company is a productive powerhouse of a multinational corporation, As with most businesses that they have a vested curiosity in estimating how their customers and the public feeling about their services that they provide and the products that they produce, but obtaining that information, whether through in-person, over the phone, or surveys, cost money and time to create, distribute, gather, and analyze. Therefore, I am interested in observing if I use Natural Language Processing (NLP) tools and unsupervised machine learning to assess public opinion of Disney at a minimal cost.
Data acquisition
First, I wanted data in the form of text, which is publicly posted; hence I turned to Twitter. I was using the Twitter API (Application Program Interface), which takes less than 24 hours to get approval. Next, Fetched 50,000 tweets related to the keyword and hashtag #Disney in English for a week.
Data cleaning :
-
Remove URLs , Stop-words , Punctuations , Numbers
-
Convert text to lower-case
-
Convert emojis to words
-
Lemmatization
Words Counts
Vectorize & Dimensionality Reduction
I have tried a bunch of techniques to reduce the dimensionality like Latent Semantic Analysis (LSA), Biterm, and Non-negative Matrix Factorization (NMF). However, I determined that NMF with a count vectorizer works more suitable for the data regarding the short length and lack of context.
Sentiments Analysis :
IT IS DISNEY !! ALL POSITIVES ! :)
