Utilizing NLP to Gain Insights from Employee Feedback

Retaining and attracting employees is an important strategic goal of high-performing organizations. Knowing this, an undisclosed client utilized employee feedback surveys to keep a pulse on engagement. While the client had a firm grasp of quantitative analytics, they recognized that they could gain more significant insights through employees’ comments. This project applied NLP to the client’s employee experience surveys. Objectives achieved included selecting the best model to predict sentiment analysis, analyzing sentiment results, identifying frequently used words, creating word clouds, and providing a CSV output of the restructured dataset. Manual sentiment analysis was performed to determine predicted sentiment accuracy. Of four untrained models (TextBlob, VADER, NLTK, and Transformer) and three machine learning models (Logistic Regression, Naïve Bayes, Transformer), the trained transformer model was found to be the best with the highest overall accuracy, far superior negative recall, and excellent positive recall. Other key findings were reported in this paper. This project closed with recommendations and future enhancements ensuring the client remains competitive.