Seasonal Affective Disorder (SAD) Dataset - 4646

We have obtained data from May 2022 to October 2023 for our suggested framework modelling. This set of data incorporates seasonality-related speech, which we convert into text, Facebook, and Twitter posts. On the whole, 4646 data elements have been acquired, comprising 3716 representing affected individuals and the remainder of 930 representing unaffected individuals, which generated a proportional 4:1 ratio. To further enhance the effectiveness of the system, we were able to employ the Synthetic Minority Over-sampling Technique (SMOTE) for balancing the dataset using oversampling, resulting in a balance between affected and unaffected classes. After we balance our data, we proceed to implement our proposed machine learning algorithms.

本框架建模所采用的数据集涵盖自2022年5月至2023年10月期间的数据。该数据集融合了与季节性相关的语音、文本以及来自Facebook和Twitter的帖子。总体而言，共收集到4646个数据元素，其中3716个代表受影响个体，其余930个代表未受影响个体，形成了4:1的比率。为进一步提升系统效能，我们得以采用合成少数类过采样技术（SMOTE）通过过采样手段平衡数据集，从而在受影响与未受影响类别之间实现了均衡。数据平衡后，我们进而实施所提出的机器学习算法。