Principles of Industrial Engineering: Forecasting Course

Lesson 1 • The Forecasting Process Model

Presents a structured, repeatable process for generating forecasts. Connects each process step to data collection, model selection, and performance review.

Lesson 2 • What Forecasting Means in Industry

Defines forecasting and distinguishes it from estimation and planning. Anchors the concept within supply chain, production, and capacity management contexts.

Lesson 3 • Taxonomy of Forecast Types

Classifies forecasts by time horizon, functional area, and data source. Students map forecast categories to real industrial use cases.

Lesson 1 • Exploratory Data Analysis for Forecasting

Applies descriptive statistics and visualization to reveal patterns before modeling. Students identify trends, seasonality, and anomalies in industrial demand data.

Lesson 2 • Cleaning and Imputation Methods

Covers practical techniques for correcting defects in raw data. Students apply mean, median, interpolation, and model-based imputation to industrial time series.

Lesson 3 • Time-Series Structuring and Indexing

Teaches proper temporal indexing and frequency alignment for time-series data. Students restructure raw logs into uniform, analysis-ready formats.

Lesson 4 • Sources of Industrial Data

Surveys internal and external data sources relevant to industrial forecasting. Students evaluate source reliability, frequency, and coverage for specific forecast goals.

Lesson 1 • Expert Opinion and Structured Interviews

Covers techniques for eliciting and structuring expert knowledge into forecast inputs. Students design interview protocols and aggregate expert estimates systematically.

Lesson 2 • Delphi Method in Industrial Settings

Applies the iterative Delphi process to achieve expert consensus on uncertain forecasts. Students facilitate a simulated Delphi round and analyze convergence.

Lesson 3 • Cognitive Biases in Judgmental Forecasting

Identifies common cognitive biases that distort expert-based forecasts. Students apply debiasing techniques to improve the accuracy of judgmental inputs.

Lesson 4 • Market Research and Scenario Analysis

Uses customer surveys and scenario planning to generate demand forecasts. Students build scenario trees and assign probability-weighted forecast values.

Lesson 1 • Classical Decomposition Methods

Applies moving-average-based classical decomposition to extract components. Students perform manual decomposition on industrial datasets and interpret results.

Lesson 2 • Interpreting Decomposition for Decisions

Translates decomposition outputs into actionable operational insights. Students connect component behavior to inventory, staffing, and production planning decisions.

Lesson 3 • STL and Advanced Decomposition

Introduces STL (Seasonal-Trend decomposition using Loess) and its advantages over classical methods. Students apply STL to datasets with complex or changing seasonality.

Lesson 4 • Components of a Time Series

Defines trend, seasonality, cyclicality, and irregular noise as distinct components. Students distinguish each component using industrial demand examples.

Lesson 1 • Holt's Double Exponential Smoothing

Extends SES to capture linear trends using a second smoothing equation. Students apply Holt's method to trended industrial demand series.

Lesson 2 • Simple Exponential Smoothing

Covers SES as an adaptive alternative to moving averages for level-stationary series. Students optimize the smoothing parameter using error minimization.

Lesson 3 • Simple Moving Averages

Introduces the simple moving average as a baseline smoothing tool. Students compute SMA forecasts and analyze the effect of window length on responsiveness.

Lesson 4 • Selecting and Evaluating Smoothing Models

Provides a decision framework for choosing among smoothing methods based on data characteristics. Students use error metrics to compare and validate model choices.

Lesson 1 • Incorporating Trend and Seasonality in Regression

Adds time-trend variables and seasonal dummy variables to regression models. Students capture temporal patterns within a regression framework.

Lesson 2 • Regression Diagnostics and Validation

Teaches residual analysis and assumption testing to ensure model validity. Students detect and correct violations of regression assumptions in forecast models.

Lesson 3 • Multiple Regression with Causal Variables

Extends regression to include multiple predictors such as price, promotions, and economic indicators. Students build and interpret multivariate demand models.

Lesson 4 • Nonlinear and Polynomial Regression

Extends regression to capture curved relationships between predictors and demand. Students fit and evaluate polynomial and log-transformed regression models.

Lesson 1 • Stationarity and Differencing

Defines stationarity and explains why it is required for ARIMA modeling. Students apply differencing and statistical tests to achieve and confirm stationarity.

Lesson 2 • ARIMA Model Identification and Estimation

Applies the Box-Jenkins identification, estimation, and diagnostic cycle. Students select ARIMA orders using information criteria and residual diagnostics.

Lesson 3 • Seasonal ARIMA Models

Extends ARIMA to handle seasonal patterns using the SARIMA framework. Students identify seasonal orders and fit SARIMA models to industrial demand data.

Lesson 4 • Forecast Generation and Uncertainty Bounds

Generates multi-step forecasts from fitted ARIMA models with prediction intervals. Students interpret interval width as a measure of forecast uncertainty.

Lesson 1 • Forecast Error Metrics

Defines and computes key error metrics used to quantify forecast accuracy. Students select appropriate metrics based on scale, outlier sensitivity, and business context.

Lesson 2 • Benchmarking and Model Comparison

Establishes naive and seasonal naive benchmarks to contextualize model performance. Students conduct structured model comparison using consistent evaluation protocols.

Lesson 3 • Validation Strategies for Forecast Models

Covers hold-out, rolling-origin, and cross-validation approaches for time-series models. Students implement rolling-origin evaluation on industrial datasets.

Lesson 4 • Continuous Forecast Improvement Process

Establishes a structured review cycle for ongoing forecast performance management. Students design a forecast review cadence and improvement action plan.