A

MINI PROJECT REPORT

On
“Multi-Class Classification of Stress using Text Data”

Submitted to
Autonomous Institute,
Affiliated to The Rashtrasant Tukadoji Maharaj Nagpur University
Department of Emerging Technologies
Bachelor of Technology (B. Tech)

Submitted By
1. Shyam Patil (AD22029)
2. Ishant Morghade (AD22040)
3. Bhavesh More (AD22050)
4. Peeyush Kaware (AD22057)

Guided By
Prof. Kalyani Pendke

S. B. JAIN INSTITUTE OF TECHNOLOGY, MANAGEMENT AND

RESEARCH, NAGPUR
2024 – 2025
 CERTIFICATE

This is to certify that the mini project report entitled “Multi-Class Classification of
Stress using Text Data” submitted by Ishant Morghade, Shyam Patil, Bhavesh
More and Peeyush Kaware to the S. B. JAIN INSTITUTE OF TECHNOLOGY,
MANAGEMENT AND RESEARCH, NAGPUR of B.Tech in (Emerging
Technologies) is a Bonafide record of mini project work carried out by him/her under
my supervision. The contents of this report, in full or in parts, have not been submitted
to any other Institution or University for the award of any degree or diploma.

<Signature> <Signature>

1. Shyam Patil AD22029 Prof. Kalyani Pendke

2. Ishant Morghade AD22040
3. Bhavesh More AD22050
4. Peeyush Kaware AD22057

Signature of HOD with seal

 DECLARATION

We declare that this mini project report titled “Multi-Class Classification of Stress
using Text Data” of B. Tech in (Emerging Technologies) is a record of original work
carried out by us under the supervision of Prof. Kalyani Pendke, and has not formed the
basis for the award of any other degree or diploma, in this or any other Institution or
University. In keeping with the ethical practice in reporting scientific information, due
acknowledgements have been made wherever the findings of others have been cited.

<Signature>

1. Shyam Patil (AD22029)

2. Ishant Morghade (AD22040)
3. Bhavesh More (AD22050)
4. Peeyush Kaware (AD22057)

<Date>
 INDEX
Sr
Topics Page No.
No.
1 Abstract 1
2 Introduction 2
3 Aims & Objectives of Project 3
4 Literature Review 4
5 Proposed Work 5
6 Research Methodology 6
7 Result 7
8 Conclusion 8
9 Future Scope 9
10 References 10
11 Bibliography 11
 ABSTRACT

Stress is a critical factor affecting mental health, and its timely detection can aid in early
intervention and psychological well-being. Traditional stress assessment methods often rely
on self-reported surveys, which may be subjective and time-consuming. This project aims to
develop an automated multi-class classification model to detect stress levels (low, moderate,
high) from textual data, including social media posts, journal entries, and chat
conversations. Our approach leverages Natural Language Processing (NLP) and machine
learning to process and analyze text data. The methodology includes data preprocessing,
feature extraction, and classification using various machine learning models and deep
learning architectures. Given the imbalanced nature of stress-related datasets to ensure
robust model performance. The expected outcome is an efficient, real-time stress detection
model capable of identifying stress patterns from text. This research has significant
applications in mental health monitoring, AI-driven therapy assistants, and workplace stress
assessment, contributing to scalable, data-driven psychological support while addressing
ethical considerations related to privacy and bias mitigation.
Chapter-1
 Chapter 1: Introduction
Stress is a growing public health concern, significantly impacting mental
and physical well-being. With the increasing use of digital communication,
people often express their emotions, including stress, through textual
sources such as social media posts, personal journals, and online chat
conversations. In an increasingly fast-paced and digitally connected world,
stress has become a ubiquitous part of daily life. Chronic stress is linked to
a myriad of mental and physical health issues, including anxiety,
depression, cardiovascular diseases, and weakened immune function.
Traditional methods of stress detection, such as self-reported surveys or
clinical assessments, are often subjective, time-consuming, and limited in
scalability. With the proliferation of digital communication platforms,
individuals frequently express their emotions and mental states through
text-based mediums such as social media posts, journal entries, and chat
conversations. This presents a unique opportunity to leverage natural
language processing (NLP) and machine learning (ML) techniques to detect
stress levels from textual data in a non-invasive and scalable manner.
The "Multi-Class Classification of Stress Detection using Text Data"
project aims to develop an automated system capable of classifying stress
levels (e.g., low, moderate, high) based on textual input. By analyzing
patterns in language use, sentiment, and emotional expression, the system
can provide insights into an individual's mental state, enabling early
intervention and support. This approach is particularly relevant in today’s
context, where mental health issues are on the rise, and there is a growing
need for accessible and effective mental health monitoring tools.
Importance of Stress Detection from Text Data
Textual data from social media, journals, and chats offers a rich source of
information about an individual's emotional and psychological state. Unlike
traditional methods, text-based stress detection can:
1. Provide Real-Time Insights: Continuous monitoring of text data allows
for real-time detection of stress levels, enabling timely interventions.
2. Offer Non-Invasive Monitoring: Unlike wearable devices or clinical
assessments, text-based analysis does not require physical interaction,
making it more accessible and less intrusive.
3. Scale Across Populations: Automated systems can analyze large volumes
of text data, making it feasible to monitor stress levels across diverse
populations.
Challenges in Analyzing Text for Mental Health
Despite its potential, analyzing text data for stress detection presents several
challenges:
1. Ambiguity and Context: Textual expressions can be ambiguous, and the
same words may convey different meanings depending on the context.
2. Sarcasm and Irony: Detecting sarcasm or irony in text is particularly
challenging, as these expressions often mask true emotions.
3. Cultural and Linguistic Variations: Expressions of stress vary across
cultures and languages, requiring models to be adaptable and inclusive.
4. Data Imbalance: Stress-related text may be underrepresented in datasets,
leading to biased or inaccurate models.
Multi-class classification is a critical component of this project, as it allows
for the categorization of stress levels into distinct classes (e.g., low,
moderate, high). This granularity is essential for providing actionable
insights and tailored interventions. For instance:
• Low Stress: May require minimal intervention, such as mindfulness
exercises.
• Moderate Stress: Could benefit from counseling or stress management
programs.
• High Stress: May necessitate immediate professional support or crisis
intervention.
By employing advanced NLP techniques and machine learning models, the
system can accurately classify stress levels while addressing the challenges
mentioned above.
Chapter-2
 Chapter 2: Aims & Objectives of Project

aim: - “multi-class classification of stress using text data”

OBJECTIVE: -
• ANALYZE WRITTEN CONTENT.
• IDENTIFY DIFFERENT CATEGORIES OF STRESS (E.G., LOW, MODERATE, HIGH).
• HELP IN UNDERSTANDING STRESS PATTERNS BASED ON TEXT ANALYSIS.
• SUPPORT MENTAL HEALTH MONITORING AND EARLY STRESS DETECTION.
Chapter-3
Chapter 3: Literature Review
Chapter-4
Chapter 4: Literature Review
Chapter-5
Chapter 5: Proposed Work
Chapter-6
Chapter 6: Research Methodology
Chapter-7
Chapter 7: Result
Chapter-8
Chapter 8: Conclusion
Chapter-9
Chapter 9: Future Scope
Chapter-10
 Chapter 10: References

[1] SARNO R, GHOZALI I. MULTI-CLASS MULTI-LEVEL CLASSIFICATION OF

MENTAL HEALTH DISORDERS BASED ON TEXTUAL DATA FROM SOCIAL
MEDIA. JOURNAL OF INFORMATION AND COMMUNICATION TECHNOLOGY.
2024 JAN 30.
[2] DANURI MS, ABD RAHMAN R, MOHAMED I, AMIN A. THE IMPROVEMENT
OF STRESS LEVEL DETECTION IN TWITTER: IMBALANCE CLASSIFICATION
USING SMOTE. IN2022 IEEE INTERNATIONAL CONFERENCE ON COMPUTING
(ICOCO) 2022 NOV 14.
[3] PIMPALKAR A, MIGLANI D, RIZVI A, DANDEKAR P, VERMA S, SELOKAR K.
ENSEMBLE LEARNING TECHNIQUES FOR CLASSIFYING STRESSED AND
UNSTRESSED TEXTUAL DATA. IN2024 IEEE INTERNATIONAL STUDENTS
CONFERENCE ON ELECTRICAL, ELECTRONICS AND COMPUTER SCIENCE
(SCEECS) 2024 FEB 24.
[4] SOWBARNIGAA KS, SHANMUGAVADIVEL K, SRIRAAM RV, SUBRAMANIAN
M, SANDHIYA R, SATHISHKUMAR VE. LEVERAGING MULTI- CLASS
SENTIMENT ANALYSIS ON SOCIAL MEDIA TEXT FOR DETECTING SIGNS OF
DEPRESSION. APPLIED AND COMPUTATIONAL ENGINEERING.
[5] BENDEBANE L, LABOUDI Z, SAIGHI A, AL-TARAWNEH H, OUANNAS A,
GRASSI G. A MULTI- CLASS DEEP LEARNING APPROACH FOR EARLY
DETECTION OF DEPRESSIVE AND ANXIETY DISORDERS USING TWITTER
DATA. ALGORITHMS. 2023 NOV 27.
[6] GARG M. MULTI-CLASS CATEGORIZATION OF REASONS BEHIND MENTAL
DISTURBANCE IN LONG TEXTS. KNOWLEDGE-BASED SYSTEMS. 2023 SEP
27;276:110742.
[7] PANKAJAVALLI PB, KARTHICK GS. AN INDEPENDENT CONSTRUCTIVE
MULTI-CLASS CLASSIFICATION ALGORITHM FOR PREDICTING THE RISK
LEVEL OF STRESS USING MULTI-MODAL DATA. ARABIAN JOURNAL FOR
SCIENCE AND ENGINEERING. 2022 AUG;47(8):10547-62.
[8] RATHOD Y, VAGHELA D. EVALUATION OF FEATURE SELECTION AND
MULTI-CLASS PREDICTION METHODS FOR METAL STRESS. IN2022
INTERNATIONAL CONFERENCE ON AUTOMATION, COMPUTING AND
RENEWABLE SYSTEMS (ICACRS) 2022 DEC 13 (PP. 1091-1095). IEEE.
[9] DRITSAS E, TRIGKA M. UTILIZING MULTI-CLASS CLASSIFICATION
METHODS FOR AUTOMATED SLEEP DISORDER PREDICTION. INFORMATION.
2024 JUL 23;15(8):426.
[10] YANG Q, LI Y, LI B, GONG Y. A NOVEL MULTI-CLASS CLASSIFICATION
MODEL FOR SCHIZOPHRENIA, BIPOLAR DISORDER AND HEALTHY CONTROLS
USING COMPREHENSIVE TRANSCRIPTOMIC DATA. COMPUTERS IN BIOLOGY
AND MEDICINE. 2022 SEP 1;148:105956.
Chapter-11
 Chapter 11: Bibliography
1. BOOKS:
1. Aggarwal, C. C. (2018). Neural Networks and Deep Learning: A Textbook. Springer.
2. Bishop, C. M. (2006). Pattern Recognition and Machine Learning. Springer.
3. Alpaydin, E. (2020). Introduction to Machine Learning (4th ed.). MIT Press.

2. JOURNAL ARTICLES:
1. Coppersmith, G., Dredze, M., Harman, C., & May, D. (2014). CrisisBERT: A
Multi-Class Text Classification Approach to Stress Detection. Proceedings of the
27th International Conference on Computational Linguistics, 1125-1134.
2. Amin, A., & Mollah, M. A. (2020). Stress Detection in Text Using Machine
Learning Approaches: A Survey. Journal of Computer Science and
Technology, 35(2), 121-130.
3. Ananthakrishna, S. (2022) : Stress Detection in Online Texts Using
Machine Learning and NLP. Journal of Big Data, 9(1), 23-45.

3. WEBSITES:

1. Wikipedia. (n.d.). Retrieved from www.wikipedia.org

2. Medium. (2021). How Machine Learning is Used to Detect Stress from Text.
Retrieved from www.medium.com
3. Towards Data Science. (2020). A Beginner's Guide to Text Classification Using Machine
Learning. Retrieved from www.towardsdatascience.com
4. Kenton, J., & Toutanova, K. (2019): BERT: Pre-training of deep bidirectional
transformers for language understanding. Retrieved from
https://arxiv.org/abs/1810.04805
The foundational paper on BERT, which is central to your project for extracting contextual
text embeddings and improving stress classification accuracy.