leev75/volz-nlp-sentiment-analysis

NLP pipeline for sentiment analysis of 527 Google Play reviews — Volz flight-booking app (Algeria). Arabic · French · English · 15 visualizations · 6 business enhancements. # Volz App — NLP Sentiment Analysis **Google Play Store · 527 Reviews · Aug 2024 – Mar 2026** A full NLP pipeline applied to real user reviews of Volz — Algeria's leading flight-booking app. The analysis transforms multilingual user feedback (Arabic, French, English) into a ranked product roadmap and actionable business insights. --- ## Project Overview | Phase | Description | |-------|-------------| | 1 | Data Collection — Google Play scraping (5 configs) | | 2 | Preprocessing & Feature Engineering | | 3 | Exploratory NLP Visualizations (9 charts) | | 4 | Business Insights & Recommendations | | 5 | Arabic NLP Pipeline (CAMeL Tools) | | 6 | App Version Sentiment Tracker | | 7 | Payment Method Granularity Analysis | | 8 | B2B / Corporate Signal Detector | | 9 | Diaspora vs Local User Segmentation | --- ## Key Findings - **74.5%** of reviews are Positive — the app is broadly liked - **Bugs & Crashes = 100% Pain Index** — every mention is a complaint - **Pricing** is the #1 driver of dissatisfaction (`trop cher`, `prix`) - **Core strengths to protect:** `rapide`, `facile`, `excellent` - **Bimodal rating pattern** — users either love (5★) or hate (1★) the app --- ## Tech Stack ``` Python · Pandas · NLTK · Matplotlib · Seaborn NetworkX · WordCloud · CAMeL Tools · google-play-scraper ``` --- ## Setup & Usage ```bash # Install dependencies pip install google-play-scraper pandas nltk matplotlib seaborn \ wordcloud networkx camel-tools scikit-learn # Run Phase 1 first to collect data # Then run each phase sequentially ``` > **Note:** Phase 1 (Data Collection) requires an internet connection to scrape Google Play. All subsequent phases read from the saved CSV file. --- ## Dataset - **Source:** Google Play Store — `app.volz` - **Size:** 527 unique reviews after deduplication - **Period:** August 2024 – March 2026 - **Languages:** Arabic, French, English - **Scraping configs:** `ar/dz · fr/dz · en/dz · fr/fr · en/us` --- ## Visualizations | Chart | Descrip …