WILSONIKENNA/AI05-Nigeria-Language-Identifier

AI-powered Nigerian language identification system built with machine learning and Streamlit. 🇳🇬 AI05 Nigeria Language Identifier 3MTT NextGen Capstone Project Author: Wilson Ikenna Ugwu AI05 Nigeria Language Identifier is a machine-learning application that identifies selected Nigerian languages from written text. The current version supports English, Igbo, Yoruba, Hausa, and Nigerian Pidgin. The trained model is integrated into a Streamlit web application that allows users to enter text, identify the most likely language, view prediction confidence, and compare the probabilities of all supported languages. 🎯 Project Objectives Identify selected Nigerian languages from written text using machine learning. Prepare and preprocess a multilingual text dataset. Train and evaluate a language-classification model. Analyze model performance using accuracy, precision, recall, F1-score, and a confusion matrix. Provide a simple web interface for real-time language identification. Display prediction confidence and language probabilities. 🌍 Supported Languages | Language | Supported | |----------|-----------| | English | ✅ | | Igbo | ✅ | | Yoruba | ✅ | | Hausa | ✅ | | Nigerian Pidgin | ✅ | 📊 Model Performance The final model achieved 99.49% overall test accuracy. Nigerian Pidgin was the most challenging class. After retraining, its recall improved from approximately 84% to 88%. Language Correct / Test Samples English 500 / 500 Hausa 259 / 261 Igbo 499 / 499 Nigerian Pidgin 43 / 49 Yoruba 272 / 272 The main remaining classification error was Nigerian Pidgin being classified as English. 🧠 How It Works Dataset ↓ Data Preprocessing ↓ Training/Test Split ↓ Model Training ↓ Model Evaluation ↓ Confusion Matrix ↓ Prediction Testing ↓ Streamlit Web Application For a new text input, the application receives the text, sends it to the trained classifier, predicts the most likely language, calculates probabilities for the supported languages, and displays the prediction and confidence. 🖥️ Web Application The application is built with Streamlit and p …