# 🧠 Client Intent Classifier

A lightweight machine learning project that classifies customer support messages into predefined intent categories such as `return`, `complaint`, `invoice inquiry`, and more.

Built with Python, `scikit-learn`, and `TfidfVectorizer` — designed for quick local predictions and easy customization for specific business needs.

---

## 📂 Categories

The model recognizes the following intent categories:
- `zwrot` – return
- `reklamacja` – complaint
- `pytanie o zamówienie` – order inquiry
- `pytanie o fakturę` – invoice inquiry
- `niezwiązane z produktem` – unrelated
- `inne pytanie` – other question

---

## 🚀 How to use

### 1. Clone the repo
```bash
git clone https://github.com/yourusername/client-intent-classifier.git
cd client-intent-classifier

2. Install dependencies

pip install -r requirements.txt

3. Train the model

python train.py

This will create two files:

• model/model.pkl – trained classifier
• model/vectorizer.pkl – text vectorizer

4. Make predictions

python predict.py

Or use classify(text) inside your code.

---

🌐 Optional: Streamlit App

You can run a simple web interface using:

---

📊 Dataset

The training dataset is available in:

data/boc_intents_dataset.csv

It contains 200+ artificial Polish-language messages balanced across 6 categories.
You can replace this with real data and re-train the model easily.

---

🧪 Example

Input:

Chciałbym otrzymać fakturę za zamówienie.

Output:

pytanie o fakturę

---

📄 License

This project is licensed under the MIT License.
See the LICENSE file for details.

Feel free to use or adapt for educational and non-commercial use.
For commercial licensing – contact the author.

---

✨ Author

Created by Maryna Dudik Junior Data Scientist & Automation Enthusiast