The suggested model outperforms both of the previous models, therefore it's important to think about what the authors can take away from it and whether they can identify any trends in the diagnosis of depression. The authors can identify the words and phrases that have the most impact on a user's choice of category by using the logistic regression model. To better illustrate what the model found as the words that contribute the most to classification, the authors produced a word cloud using these terms. The larger the term, the bigger the donation. It is clear from this unigram word cloud that the authors would quickly link certain terms to depressed people, therefore it will be interesting to see how the model arranges words for both classes. To see how closely bigram concepts match human intuition, the authors may also reevaluate them. This data supports the idea that people who are depressed focus far more on themselves than on others.