Practical AI for Demand Forecasting and Inventory Optimization Course

Lesson 1 • What Is Demand Forecasting

Defines demand forecasting and its role in supply chain decisions. Establishes vocabulary used throughout the course.

Lesson 2 • Demand Data Preprocessing

Covers cleaning, transforming, and structuring raw demand data. Ensures students can produce model-ready datasets.

Lesson 3 • Exploratory Data Analysis for Demand

Applies EDA techniques to uncover demand patterns before modeling. Prepares students to detect anomalies and seasonality early.

Lesson 1 • ARIMA and Seasonal ARIMA

Teaches ARIMA model identification, estimation, and diagnostics. Extends to SARIMA for seasonal demand patterns.

Lesson 2 • Moving Averages and Smoothing

Introduces simple and weighted moving averages as entry-level forecasting tools. Builds intuition for lag and smoothing trade-offs.

Lesson 3 • Exponential Smoothing Models

Covers single, double, and triple exponential smoothing for trend and seasonal data. Connects smoothing parameters to forecast responsiveness.

Lesson 4 • Forecast Accuracy Metrics

Defines key error metrics and explains when each is appropriate. Enables objective model comparison across methods.

Lesson 1 • Framing Forecasting as ML

Reframes time series forecasting as a supervised learning problem. Bridges classical methods to ML workflows.

Lesson 2 • Regression and Linear Models

Applies regularized regression to demand forecasting with many features. Provides interpretable alternatives to complex models.

Lesson 3 • Model Evaluation and Selection

Teaches time-series-aware cross-validation and model selection protocols. Ensures robust generalization to future demand periods.

Lesson 4 • Tree-Based Forecasting Models

Covers gradient boosting and random forest models for demand prediction. Highlights their strength with tabular, multi-feature datasets.

Lesson 1 • Training and Tuning Deep Models

Covers practical training workflows including learning rate scheduling and early stopping. Reduces trial-and-error in deep model development.

Lesson 2 • Transformer-Based Forecasting

Introduces attention mechanisms and Transformer models for long-horizon forecasting. Covers efficient variants suited to demand time series.

Lesson 3 • Neural Network Fundamentals

Reviews feedforward networks and backpropagation as prerequisites for sequence models. Grounds deep learning concepts in forecasting context.

Lesson 4 • Convolutional Models for Time Series

Applies 1D CNNs to extract local temporal patterns in demand data. Offers a faster alternative to recurrent architectures.

Lesson 1 • Probabilistic Forecast Fundamentals

Distinguishes point forecasts from probabilistic forecasts and their business value. Introduces quantile and interval forecasting concepts.

Lesson 2 • Quantile Regression and Conformal Methods

Applies quantile regression and conformal prediction to generate calibrated intervals. Enables risk-aware inventory decisions.

Lesson 3 • Bayesian Forecasting Approaches

Uses Bayesian structural time series and probabilistic programming for demand. Provides principled uncertainty quantification.

Lesson 4 • Hierarchical Demand Structures

Models demand across SKU, category, and regional hierarchies. Ensures forecasts are consistent across aggregation levels.

Lesson 1 • Economic Order Quantity Models

Derives EOQ and its extensions for quantity discounts and backorders. Provides closed-form solutions for order sizing.

Lesson 2 • Multi-Echelon Inventory Basics

Extends single-location models to multi-echelon supply networks. Introduces stock positioning across distribution tiers.

Lesson 3 • Safety Stock and Reorder Points

Calculates safety stock using demand variability and lead time uncertainty. Links forecast error directly to inventory buffer sizing.

Lesson 4 • Continuous vs. Periodic Review

Compares continuous and periodic inventory review systems and their cost implications. Guides policy selection based on operational constraints.

Lesson 1 • Inventory Segmentation with ML

Clusters SKUs by demand pattern and applies differentiated policies per segment. Reduces complexity while improving overall inventory performance.

Lesson 2 • Linking Forecasts to Inventory Decisions

Translates probabilistic forecasts into dynamic safety stock and reorder parameters. Closes the loop between forecasting and inventory policy.

Lesson 3 • Newsvendor and Stochastic Models

Applies the newsvendor model to single-period and perishable inventory problems. Extends to stochastic dynamic programming for multi-period decisions.

Lesson 4 • Reinforcement Learning for Replenishment

Frames inventory replenishment as a Markov decision process solved with RL. Enables policies that learn from simulated supply chain environments.

Lesson 1 • Forecast Governance and Bias Auditing

Establishes processes to audit systematic forecast bias and ensure accountability. Aligns AI outputs with business and ethical standards.

Lesson 2 • Measuring Business Impact

Quantifies inventory savings, service level gains, and ROI from AI forecasting. Connects technical metrics to executive-level business outcomes.

Lesson 3 • Model Deployment and Serving

Packages and serves forecasting models via APIs and batch jobs. Covers containerization and scalable inference patterns.

Lesson 4 • Monitoring and Drift Detection

Tracks forecast accuracy and data drift in production environments. Triggers retraining when model performance degrades.