In the most recent period, the discipline of Analyzing the emotions expressed on Twitter has grown rapidly, with several studies supporting the utilization of algorithms based on machine learning techniques to analyze tweets and extract user sentiments about a certain subject. This work intends to do a comprehensive analysis of the emotional tone of tweets by making use of ordinal regression as well as other machine learning approaches. Following the completion of the preprocessing of the tweets, the suggested technique implements a method for the extraction of features in order to generate a reliable feature. After that, a number of different criteria are used in order to rank and assign weights to these attributes. The emotional states can be detected through the utilization of several techniques, such as multinomial logistic regression using SoftMax, support vector regression (SVR), decision trees (DTs), and random forests (RF).The software makes advantage of NLTK corpora resources that are open to the public, namely a Twitter dataset. The experimental findings indicate that the proposed approach for recognizing ordinal regression using methods derived from machine learning is accurate to a high degree. Additionally, it would seem that Decision Trees performs better than any other algorithm that was tested.