🎫 Support Ticket Classification using Machine Learning

A Machine Learning project that automatically classifies customer support tickets into different categories and predicts their priority level using Natural Language Processing (NLP).

This project was developed as part of Future Interns – Machine Learning Task 2.

---

🚀 Live Demo

🌐 Streamlit App

https://saktronx-ticket-classifier.streamlit.app/

---

📌 Project Overview

Customer support teams receive thousands of tickets every day. Manually identifying the ticket category and priority is time-consuming and inefficient.

This project automates that process using Machine Learning by predicting:

• 🎫 Ticket Category
• 🚨 Ticket Priority

The application is deployed using Streamlit and allows users to enter any support ticket to receive instant predictions.

---

🎯 Features

• Automatic Ticket Classification
• Ticket Priority Prediction
• Text Preprocessing using NLP
• TF-IDF Feature Extraction
• Multiple ML Model Comparison
• Interactive Streamlit Web App
• Confusion Matrix Visualization
• Word Cloud Visualization
• Deployed Online

---

🧠 Machine Learning Workflow

1. Data Collection
2. Data Cleaning
3. Text Preprocessing
   • Lowercasing
   • Removing punctuation
   • Removing stopwords
   • Lemmatization
4. Feature Extraction using TF-IDF
5. Train-Test Split
6. Model Training
7. Model Evaluation
8. Model Comparison
9. Model Saving using Joblib
10. Streamlit Deployment

---

🤖 Machine Learning Models Used

• Naive Bayes
• Logistic Regression
• Linear Support Vector Machine (Linear SVM)

Among all models, Linear SVM achieved the best performance and was selected for deployment.

---

🛠️ Tech Stack

• Python
• Scikit-learn
• Pandas
• NumPy
• NLTK
• Matplotlib
• WordCloud
• Joblib
• Streamlit

---

📸 Project Screenshots

🏠 Home Page

The landing page of the Streamlit application.

---

🎫 Prediction Example 1

Predicting ticket category and priority for a technical support issue.

---

🎫 Prediction Example 2

Prediction for another customer support ticket.

---

📊 Model Comparison

Comparison of classification accuracy across different machine learning models.

---

📈 Confusion Matrix

Evaluation of the final Linear SVM model.

---

☁️ Word Cloud

Most frequently occurring words after preprocessing.

---

📂 Project Structure

Support-Ticket-Classification/
│
├── Dataset/
│
├── models/
│   ├── classifier.pkl
│   └── vectorizer.pkl
│
├── notebook/
│
├── screenshots/
│   ├── home.png
│   ├── prediction1.png
│   ├── prediction2.png
│   ├── model_comparison.png
│   ├── confusion_matrix.png
│   └── wordcloud.png
│
├── app.py
├── requirements.txt
└── README.md

---

⚙️ Installation

Clone the repository

git clone https://github.com/saktronX/FUTURE_ML_02.git

Move into the project folder

cd FUTURE_ML_02

Install dependencies

pip install -r requirements.txt

Run the Streamlit application

streamlit run app.py

---

📊 Dataset

The dataset contains customer support tickets with the following information:

• Ticket Text
• Ticket Category
• Ticket Priority

Example categories include:

• Technical Issue
• Billing Inquiry
• Product Inquiry
• Cancellation Request
• Refund Request

Priority levels include:

• Low
• Medium
• High
• Critical

---

📈 Results

• Successfully trained multiple ML models.
• Compared model performance.
• Selected Linear SVM for deployment.
• Built a fully functional Streamlit application.
• Predicted both ticket category and priority.
• Successfully deployed online.

---

🔮 Future Improvements

• Deep Learning models (LSTM/BERT)
• Confidence score for predictions
• Multi-language ticket classification
• Automatic ticket routing
• Email integration
• Cloud database support

---

👨‍💻 Author

Saksham Verma

GitHub: https://github.com/saktronX

---

📜 License

This project is developed for educational purposes as part of the Future Interns Machine Learning Internship Program.