This article identified a lack in scientific literature for topics such as sentiment analysis systems, pipelines, and teacher evaluation. I exposed the challenges in developing sentiment analysis for teaching performance. An Anaconda environment with Jupyter Notebook and Spyder IDE was used to build the machine learning pipeline. To implement the pipeline, I used a large (2 million records and 8 attributes), cleaned, open-source data set for aspect-based sentiment analysis. The focus was to prepare a reduced pandas DataFrame by filtering the larger data set by the number of total words for student comment, sentiment type, and subjectivity type. The word cloud feature provides a clear status for teaching performance in visual approach. To produce sets of advice for each student comment, I have integrated the OpenAI and MistralAI models, comparing the generated content from both approaches in terms of LLM size, price, quality of content, and the execution time.