Finding the most prevalent and pertinent user opinions on a certain topic is key to the effectiveness of sentimental analysis. In this study, a framework is created combining data pre-processing techniques, a blend of machine learning, statistical modelling, and deep learning methodologies, to assess user sentiments on Twitter regarding natural catastrophes. When using text data for natural language processing, features can be retrieved piecemeal and compared, but without taking context or the entire sentence into account, the sentiment may be misunderstood. This research suggests utilising a combination of deep learning and machine learning models to evaluate pre-labelled Twitter feeds on disasters to understand the feelings surrounding a certain tragedy. In our study, we glean information from tweets about various disasters that reflect the emotions expressed during the tragic events as reported by Twitteratti. To use machine learning and deep learning algorithms, data is cleaned, purified, and optimised. We provide several machine learning and deep learning models and assess their performances on many metrics in addition to exploratory data analysis.