Machine Learning Engineering Projects in PunjabMachine Learning Engineering Projects in Punjab
Cover image for Predictive Analytics Dashboard
Dataset: DataCo Smart
Predictive Analytics Dashboard Dataset: DataCo Smart Supply Chain Dataset Size: ~180,519 Rows | 53 Columns 🛠️ Phase 1: Tech Stack & Tools Used This project was completed using industry-standard tools: Data Manipulation: Pandas, NumPy Visualization: Matplotlib, Seaborn (for static EDA) and Plotly Express (for interactive dashboards) Machine Learning: Scikit-learn (for preprocessing and metrics), XGBoost (for advanced regression) Forecasting: Facebook Prophet (for seasonality) and XGBoost (for demand volume prediction) Deployment: Streamlit (to build the live BI dashboard) 🧹 Phase 2: Major Hurdles & Data Cleaning The dataset was heavily corrupted, presenting several challenges: Misplaced Data: City names like “São Paulo,” “Rio de Janeiro,” and “Grande del Norte” were incorrectly placed in the Order Status column. Missing Statuses: Many rows had empty Order Status fields, while the actual status was found in Order State. Solution: I developed a custom Restoration Engine that cleaned columns and relocated misplaced data to their correct fields (Order Region). 💰 Phase 3: Profit & Strategy Analysis Beyond visualization, the project supported business decision-making: Profit Analysis: Calculated profit margins for each product. Price Optimization: Suggested a 5% price increase for high-selling products with margins below 10% to improve profitability. 🚀 Phase 4: Modeling & Predictions (Next 5 Months) Advanced XGBoost models were used instead of basic regressions: Demand Forecast: Predicted order volumes for the next five months to enhance inventory management. Sales Trends: Optimized models to capture seasonality and trend effects on future sales. 🏭 Phase 5: The Final BI Dashboard Built a complete interactive system using Streamlit: 11 Industry-Level Visualizations: Demand trends, top-selling products, regional sales, and late delivery root causes. Interactive System: Management can use live filters to extract insights from over 100,000 rows of data. 🧠 I’ll help you extract business insights through data analysis, dashboards, and forecasting. Here’s my portfolio: kazimhaidersyedportfolio.lovable.app (http://kazimhaidersyedportfolio.lovable.app)
0
27