This study analyzed the intensity of parenting-related complaints posted in Japanese. Complaints are defined as expressions of dissatisfaction that do not aim to seek improvement from the target. Research on complaints in Japanese is few, and the factors contributing to the severity of complaints have not been thoroughly analyzed. To address this gap, Best-Worst Scaling (BWS) was utilized to annotate complaint intensity. BWS is a method that enables quantitative evaluation of elements that are difficult to quantify by repeatedly selecting the strongest and weakest options from multiple choices. By enabling quantitative evaluation, it becomes possible to identify whether a post represents a severe complaint. Subsequently, regression analysis was performed using various vectorization methods and pre-trained models. The results showed that regression analysis using LightGBM, with vectorization based on the Bag-of-Words method, yielded the best performance. In the discussion, an analysis of the words associated with intense complaints revealed that posts containing terms such as “Covid-19” and “Phone” were more likely to represent strong complaints. Additionally, error analysis was conducted to identify the types of posts that were frequently misclassified. It was found that posts with a high frequency of emojis or those lacking specific episodes but still conveying the writer’s emotions tended to be challenging for the model to analyze accurately.