🔐 Academic Stress Detector - Model Service

💡 Overview

Model Service of this application is splitted in two parts, first the api which was made in FastAPI. Having in mind that api was mode in Python we needed to create model also in Python specificly in pytorch and pytorch lightning. Model is basic neural network with sigmoid activation functions and couple of dropout and linear layers, which allows us to predict student stress level based on it's current data.

🪾 Branches

• /api/model/predict prediction branch

  body: {
      headache: int
      sleepQuality: int
      breathingProblems: int
      noiseLevel: int
      livingConditions: int
      safety: int
      basicNeeds: int
      academicPerformance: int
      studyLoad: int
      teacherStudentRelationship: int
      futureCareerConcerns: int
      socialSupport: int
      peerPressure: int
      extracurricularActivities: int
      bullying: int
  }
  

🎯 Model Training

Model was trained, with such hyperparameters:

learning_rate = 1e-4
dropout = 0.5
gamma = 0.995
num_epochs = 100
hidden_size = 128
num_classes = 3

🗒️ Features

• Sigmoid activation function;
• Cross entropy loss;
• Adam optimizer;
• Exponential learning rate scheduler;
• Training progress bar;
• Tensorboard integration;
• pydantic scheme validation;

⚙️ Command Tools

To work with this project locally or in a containerized environment, use the following commands:

./mvnw spring-boot:run # to run project

./mvnw clean # clean logs


export {PROPERTY_NAME}={VALUE} # set env variable 

🧠 Tech Stack