Welcome to my Data Analytics and Machine Learning Portfolio, a collection of applied projects spanning Machine Learning, Deep Learning, Business Intelligence (Power BI), Python Automation, and SQL Analytics.
Each project demonstrates hands-on experience in building real-world analytical solutions, from AI-driven forecasting to enterprise dashboards and automation pipelines.
A series of professional Power BI dashboards built to visualise and interpret operational performance, sales insights, and customer behaviour across industries.
- Comprehensive performance dashboard for multi-store liquor retail operations.
- Metrics include revenue trends, store ranking, supplier performance, and category contribution.
- Visual dashboard for a fast-food franchise analysing order trends, top products, and customer patterns.
- Includes profit margins, delivery time KPIs, and regional comparison visuals.
- Financial and customer analytics report for a simulated banking dataset.
- KPIs include deposit growth, loan performance, and customer churn trends.
SQL-based analytical models for data warehousing, reporting, and business performance analysis.
- End-to-end data model for retail liquor operations.
- Features sales, purchase orders, stock transfer, and supplier performance queries.
- Demonstrates ETL, view creation, and KPI logic using T-SQL.
- SQL schema for analysing inventory, procurement, and category-level performance in office supply chains.
- Includes advanced queries for reorder logic, seasonal trend detection, and supplier efficiency tracking.
Automation scripts designed to enhance workflow efficiency and data consistency across business systems.
- Email-driven automation system for extracting, merging, and appending daily reports.
- Automatically updates sales, stock-on-hand, and product master datasets.
- Automates collection and analysis of customer feedback.
- Uses natural language processing (NLP) to classify and summarise sentiments from business reviews.
These projects focus on AI and predictive modeling, combining computer vision, forecasting, and time-series analysis to solve business challenges.
- Goal: Classify facial emotions (Happy, Sad, Anger, Fear, Pain, Disgust) using custom CNN architectures.
- Highlights: Data augmentation, model capacity tuning, cross-domain testing, and few-shot adaptation.
- Tech Stack: PyTorch, NumPy, Pandas, scikit-learn, Matplotlib, Seaborn.
- Goal: Automate weekly product-level demand forecasting using Prophet and SARIMAX.
- Highlights: Time-series cleaning, hybrid modeling, MAE/MAPE comparison, and export-ready results.
- Tech Stack: Python, Pandas, Prophet, Statsmodels, openpyxl, scikit-learn.
📂 Machine Learning and Deep Learning Projects
├── Deep Learning for Sentiment Image Classification
└── Multi-Model Product Demand Forecasting
📂 Power BI Projects
├── Power BI - Liquor Business Analysis
├── Power BI - Pizza Business Analysis
└── Power BI - Wisabi Bank Analysis
📂 Python Projects
├── Python - Automation Data Updating
└── Python - Business Review Analysis
📂 SQL Projects
├── SQL - Liquor Business Analysis (F&B)
└── SQL - Stationary Business Analysis Office Consumables
| Category | Key Skills |
|---|---|
| Machine Learning | Model design, data preprocessing, evaluation metrics |
| Deep Learning | CNNs, PyTorch, transfer learning, augmentation |
| Forecasting | Prophet, SARIMAX, time-series decomposition |
| Power BI | DAX, data modeling, KPI dashboards, storytelling |
| SQL | Data normalization, ETL, joins, aggregation logic |
| Python Automation | Python scripting, file handling, IMAP, Excel integration |
More projects will be added in the future, expanding into:
- NLP and Large Language Model (LLM) applications
- Reinforcement Learning for pricing optimisation
- Power BI financial forecasting dashboards
- Cloud-based automation (Azure Functions / AWS Lambda)
Author: [Frank Dinh]
Email: [dinh.qnhat@gmail.com]
Year: 2025
© 2025 All Rights Reserved.
This repository is continuously updated with new applied AI and analytics projects.