Depression Detection from Reddit Text using Machine Learning
Journal of Integrated Engineering Sciences · 20.04.2026 · 2 (2) · 17-31
DOI: 10.63686/jies.v2i2034 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Integrated Engineering Sciences, online erschienen am 20.04.2026. Verfasst von Rudransh Agarwal. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
In the present paper, the focus is on various machine learning models used for depression detection on text data posted on Reddit and a comparative, in-depth analysis. In recent years, with people expressing their emotions, thoughts and mental states more often, social media is a great platform to consume user gen-erated content. These textual expressions could help appreciate human mental state and could be utilized at the initial level for identifying mental ailments, such as depression. In this study, a technique called ”Term Frequency-Inverse Document Frequency (TF-IDF)” is used to convert textual data into meaningful numerical representations. Through this change it is made possible to analyze the language use patterns and identify key information about the occurrence of depressive behaviour along with the machine learning algorithms. Four popular classification models (Logistic Regression, Naive Bayes, Support Vector Machine (SVM), and Random Forest) are implemented and evaluated with the same fixed random seed, which is also used in experiments, for repeatability and on a labeled dataset with 7,731 posts from Reddit with the same evaluation measures (accu-racy, precision, recall, F1-score and ROC-AUC). The experimental results show that the accuracy of Random Forest is 96.51 that of Logistic Regression is 95.99, that of SVM is 98 and that of Naive Bayes is 92.57 with ROC-AUC value ranges from 0.980 to 0.988. From the results these findings demonstrate the fact that these works have a strong classification power, and it can be inferred here that the following patterns of text - depression and non-depression are separa-ble. Overall these results suggest that classical machine learning and engineering measures to obtain a vector representation of the data, namely TF-IDF are valu-able and efficient for building accurate and computationally feasible depression recognition systems. The present research highlights the potential of the auto-mated analysis systems of texts in early screening of mental health conditions and in becoming part of mass mental health care solutions.
Zitieren
APA
Agarwal, R. (2026). Depression Detection from Reddit Text using Machine Learning. Journal of Integrated Engineering Sciences, 2(2), 17-31. https://doi.org/10.63686/jies.v2i2034
Vancouver
Agarwal R. Depression Detection from Reddit Text using Machine Learning. Journal of Integrated Engineering Sciences. 2026;2(2):17-31. doi:10.63686/jies.v2i2034
BibTeX
@article{Agarwal2026,
author = {Agarwal, Rudransh},
title = {Depression Detection from Reddit Text using Machine Learning},
journal = {Journal of Integrated Engineering Sciences},
year = {2026},
volume = {2},
number = {2},
pages = {17--31},
doi = {10.63686/jies.v2i2034}
}
EUVIMED