A Deep Dive into Forecasting- Excel & R.

DEMAND FORECASTING · TIME SERIES · ARIMA · HOLT-WINTERS · R PROGRAMMING · FABLE · HIERARCHICAL FORECASTING · TIDYMODELS · SUPPLY CHAIN · BUSINESS FORECASTING

★ Top Performer on Udemy for Business — Chosen by Companies for Forecasting Team Training

This course is consistently selected by companies through the Udemy for Business catalogue for training their planning, finance, supply chain, and data analytics teams. With 18,600+ enrolled students and a sustained 4.6-star rating, it is the R forecasting course that corporate learning teams trust — because its depth, real-world cases, and Excel-first pedagogy make it immediately applicable on the job.

★ Built by a Ph.D. in Forecasting — Not a Generic Data Science Instructor

Haytham’s doctoral research at the University of Bordeaux was specifically in forecasting and supply chain planning. This is not a data scientist who added a forecasting module to a general course. Every method — decomposition, ARIMA, Holt-Winters, Fable, hierarchical reconciliation — is taught by someone who spent years researching it and then deployed it with real clients across retail, supply chain, and finance. The depth shows.

★ Forecast 100,000 Time Series at Once — No Other R Forecasting Course Does This

Most forecasting courses teach you to model one time series. This course ends with you using the Fable package in R to run statistical and machine learning forecasting workflows across 10,000 to 100,000 time series simultaneously — with hierarchical aggregation, bottom-up, middle-out, and top-down reconciliation using the Minimum Trace method. This is what industrial demand planning actually looks like at scale.

★ Consistently Rated 4.6 Stars — “By Far the Best R Forecasting Course Out There”

This course has maintained a rating above 4.5 stars throughout its lifetime on Udemy — not as a launch metric, but as a sustained signal of quality from students who completed it. Verified reviewers call it “the best ever” and “by far the best R forecasting course out there.” 4.6 stars across 200 ratings and 18,600 students is not luck. It is a product of the depth, clarity, and real-world relevance of the content.

★ Excel First, R Second — Every Method Understood Before It Is Coded

This course follows a strict pedagogical rule: every statistical concept is explained and applied in Excel before a single line of R is written. Decomposition, exponential smoothing, ARIMA calibration, regression forecasting — you understand why the method works in Excel, then you scale and generalise it in R. This is why the course works for planners, analysts, and managers who are not data scientists — it meets them where they are and takes them further than they expected.

COURSE DESCRIPTION

A good forecast does not just improve accuracy — it saves resources, reduces waste, and in supply chain, actually contributes to sustainability. Forecasting is the stepping stone of every planning decision in your organisation. Get it right and everything downstream — inventory, production, procurement, finance — becomes more efficient. Get it wrong and the costs multiply through the entire value chain. This course is a complete, deep-dive program that gives you the statistical tools to get it right.

The course follows the same proven progression as all programs in this series: Excel first, R second. Every method — decomposition, smoothing, ARIMA, Holt-Winters, regression, Fable workflows — is explained and applied in Excel before a single line of R is written. R is introduced with a full crash course and then used to scale, automate, and extend every concept. By the end of the course, you will run statistical and machine learning forecasting models across 100,000 time series simultaneously using the Fable package — with hierarchical aggregation, bottom-up, middle-out, and top-down reconciliation using the Minimum Trace method.

This course is taught by a Ph.D. in forecasting — Haytham’s doctoral research at the University of Bordeaux was specifically in forecasting methodology and supply chain planning. It is a top performer on Udemy for Business, consistently rated 4.6 stars across 18,600+ students. Machine learning forecasting with Tidymodels was added by popular demand in August 2023.

Whether you are a planner who wants to understand the “why” behind every method, a data scientist who wants a supply chain forecasting framework, or a manager who wants forecasts that hold up under scrutiny — this course will take you from zero to hero.

WHAT MAKES THIS COURSE DIFFERENT

[ DEEP ]

20.5 hours of real statistical depth

This is not an overview. It is a complete forecasting program covering every statistical method from naive to hierarchical ML reconciliation — in both Excel and R, with assignments at every stage.

[ SCALE ]

100,000 time series at once with Fable

Most courses stop at one series. This one ends with you running multi-model forecasting workflows across your entire product or customer base using the Fable package and hierarchical reconciliation.

[ PHD ]

Taught by a Ph.D. in forecasting

Not a data scientist who teaches forecasting on the side. Haytham’s doctoral research was specifically in forecasting. Every method is taught with the understanding of someone who has researched and deployed it professionally.

TOOLS COVERED IN THIS COURSE

Microsoft Excel | R / RStudio | Fable | Tidymodels | Lubridate | Forecast package

WHAT YOU WILL LEARN

✓ Understand the statistical foundations of forecasting: stationarity, time series structure, causal methods, and when each approach applies

✓ Perform univariate and bivariate statistical analysis and calculate autocorrelation to understand the structure of any time series

✓ Apply and compare simple forecasting methods: naive, seasonal naive, mean, SES, weighted moving average — and select the best using MASE and MPE

✓ Decompose time series additively and multiplicatively to separate trend, seasonality, and residual components in Excel and R

✓ Build exponential smoothing models: SES, Holt, additive and multiplicative Holt-Winters with 12-month ahead forecasting

✓ Apply linear and multiple regression for forecasting in Excel and R, with full parameter calibration and accuracy measurement

✓ Learn R from scratch: installation, data structures, loops, functions, date parsing with Lubridate — a complete crash course included

✓ Fit ARIMA models in R: identify orders, train/test split, dynamic harmonic regression, grid search with SARIMA, battle of ARIMAs

✓ Use the Fable package for multi-model workflows: fit, compare, and select from statistical models including Prophet and VAR in two lines of code

✓ Apply hierarchical time series with bottom-up, middle-out, top-down, and Minimum Trace reconciliation across 10,000+ time series

✓ Build machine learning forecasting pipelines with Tidymodels: feature engineering, cross-validation, model stacking, and future prediction

✓ Forecast 100,000 time series simultaneously with accuracy measurement at every level of the hierarchy

COURSE CONTENT — 14 SECTIONS · 185 LECTURES · 20.5 HOURS

SECTION 1: Forecasting fundamentals and applications

Why does forecasting matter — and why is it hard? Understand the nature of time series data, the difference between causal and statistical methods, the concept of stationarity, and the real-world applications of forecasting across supply chain, inventory, finance, and operations. Includes the 2020/COVID case as an illustration of forecast failure and resilience. A graded quiz closes the section.

Concepts Excel

SECTION 2

Univariate and bivariate statistical analysis

Before you forecast, you must understand your data. Calculate and interpret univariate statistics (mean, variance, skewness) and bivariate relationships. Measure autocorrelation — the single most important diagnostic in time series analysis — and understand what it tells you about the structure of your series. Graded assignment with solution.

Excel R

SECTION 3: Simple forecasting methods and accuracy measurement

Build and evaluate the foundational forecasting methods: naive, seasonal naive, mean, seasonal average, simple exponential smoothing, log transformations, custom weighted moving average, and linear regression. Optimise parameters, compare methods using MASE and MPE, and identify the best simple method for your data. Graded assignment with solution.

Excel

SECTION 4: Time series decomposition

Break any time series into its structural components: trend, seasonality, and residual. Apply centred and double moving averages for detrending. Build additive and multiplicative decomposition models in Excel. Understand when each decomposition type applies and how decomposition feeds into more advanced forecasting methods. Graded assignment included.

Excel

SECTION 5

Exponential smoothing: Holt and Holt-Winters

The workhorses of operational forecasting. Build Simple Exponential Smoothing, Holt’s trend model, and the full additive and multiplicative Holt-Winters models in Excel. Initialise alpha and beta parameters, generate 12-month-ahead forecasts, and evaluate accuracy. Graded assignment with full solution.

Excel

SECTION 6

Regression forecasting in Excel and introduction to R

Apply linear and multiple regression to forecasting in Excel — fitting models, measuring accuracy, and calibrating parameters. Then make the transition to R: install R and RStudio, configure your project, install packages, and run a walkthrough tutorial. The gateway section between Excel and R.

Excel R RStudio

SECTION 7

R programming fundamentals

Learn R from absolute scratch with a forecasting mindset. Data structures and types, arithmetic and vectors, lists, dataframes, data import, selection and filtering, if-else logic, conditions, functions, for loops, and applying functions across dataframes. Includes a two-part graded assignment on real data.

R

SECTION 8

Working with dates and time in R

Dates are the foundation of every time series. Parse dates in R, extract time components, work with Lubridate for date arithmetic, model customer inter-arrival times, and build time-aware forecasting datasets. Graded assignment across six questions on real time series data.

R Lubridate

SECTION 9: Regression forecasting in R

Translate the Excel regression models into R. Prepare data for regression, handle date formatting, fit single and multiple regression forecasting models, generate predictions, measure accuracy, and evaluate model fit. Includes a two-part graded assignment.

R

SECTION 10: Time series analysis and ARIMA in R

The most technically rigorous section of the classical methods. Convert data to time series objects, analyse weekly and daily series, decompose in R, measure trend and seasonality strength, fit exponential smoothing models, identify ARIMA orders, build train/test splits, apply dynamic harmonic regression, run grid search with SARIMA using the Forecast package, and battle-test competing ARIMA models on accuracy. Graded two-part assignment.

R Forecast package

SECTION 11: Fable package: multi-model forecasting workflows

The Fable package transforms R forecasting. Build tsibble objects, calculate ACF, apply time series decomposition with Fable, use double moving average, fit multiple statistical models in a single workflow, generate and compare forecasts from linear and non-linear models, apply the Prophet model and VAR models in R, and run accuracy testing across model families. Graded Fable assignment across multiple sub-tasks.

R Fable

SECTION 12

Hierarchical and aggregate forecasting with reconciliation

Scale forecasting to the full product and channel hierarchy. Build hierarchical and grouped time series structures with tsibble. Apply aggregation: crossing levels, manual aggregations, bottom-up, middle-out, and top-down approaches. Use the Minimum Trace reconciliation method to ensure consistency across all hierarchy levels. Generate forecasts for multiple years and measure accuracy at every level simultaneously.

R Fable

SECTION 13: Machine learning forecasting with Tidymodels

Added by student demand (August 2023). Apply the Tidymodels framework to multi-level time series forecasting in R. Convert data to tsibble, generate time series features, handle missing data by level, split and log-transform data, build recipes, define model workflows, resample with cross-validation, collect and compare metrics, stack models for improved accuracy, and generate and visualise final future predictions.

R Tidymodels

SECTION 14: Multiple time series and large-scale forecasting

Put everything together at industrial scale. Fit multiple statistical models across thousands of time series simultaneously using Fable. Measure accuracy at every level of the hierarchy. Apply the full workflow — from data preparation through model selection, reconciliation, and future prediction — to a dataset with 10,000 to 100,000 time series. This is what modern, data-driven forecasting looks like in practice.

R Fable Tidymodels

THIS COURSE IS NOT FOR YOU IF...

✗ You want a Python forecasting course — this course uses R throughout. A Python version (Demand Forecasting with Excel & Python: Zero to Hero) is available separately

✗ You are looking for a 1–2 hour forecasting overview — this is a 20.5-hour complete program; it rewards sustained engagement over several weeks

✗ You want a deep learning or neural network forecasting course — this course covers statistical methods and Tidymodels ML; LSTM/transformer architectures are a separate specialisation

✗ You only need one specific method — this course is a complete forecasting program; if you need only ARIMA or only Holt-Winters, targeted shorter courses may be more appropriate

WHAT STUDENTS AND CLIENTS SAY

“By far the best R forecasting course out there. Informative and thorough. A must for aspiring Data Scientists.”

Verified student — Udemy platform

“What a course! The best ever. The depth of the content and the way Haytham builds from Excel into R is exactly how it should be taught.”

Christian — Verified Udemy student

“Excellent so far — Haytham builds every concept step by step and the way he moves from Excel into R is exactly the right approach for someone coming from a planning background.”

Babar — Verified Udemy student

“I participated in the Supply Chain Forecasting & Management training conducted by Haytham. It helped me enormously in my daily work. Haytham has the pedagogy to explain very difficult calculations and formulas in a simple way. I highly recommend this training.”

Djamel Bouremiz — Purchasing Manager, Mineral Circles Bearings W.L.L.

WHO THIS COURSE IS FOR

Demand planners and supply chain analysts

You produce forecasts manually or with basic Excel methods and want to move to validated statistical models — ARIMA, Holt-Winters, hierarchical aggregation — that you can explain and defend to stakeholders.

Finance and budget forecasting professionals

You model revenues, budgets, and P&L projections and want a rigorous, data-driven toolkit in R that goes far beyond trend lines — up to and including machine learning with Tidymodels.

Operations and inventory managers

Your forecasts drive stock levels, production plans, and procurement orders. You want the statistical foundation to make them more accurate, and the R tools to automate them across your full product range.

Retail and merchandise planners

You manage seasonal assortments across hundreds of SKUs. You want hierarchical forecasting that generates consistent, reconciled predictions at every level — category, brand, and SKU simultaneously.

Data scientists and R users entering forecasting

You know R but have not applied it to time series. You want a structured, complete forecasting curriculum — from statistical fundamentals through ARIMA, Fable, and Tidymodels ML — built around real business problems.

Students and economists studying forecasting

You are studying forecasting academically and want a course that connects statistical theory to real-world application in R — taught by a Ph.D. whose research was specifically in forecasting methodology.

REQUIREMENTS

● No prior knowledge required — the requirements field on the Udemy page reads “Nop” intentionally. This course truly starts from zero: no R, no statistics, no forecasting background assumed.

● Basic familiarity with Excel is helpful for Sections 1–6 — but Excel is used for intuition-building, not advanced analysis, so no expert level is needed.

● A computer capable of running R and RStudio (both free) — full installation walkthrough is provided in Section 6.

● The Fable, Forecast, Lubridate, and Tidymodels packages are all free and open-source — installation is guided within the relevant sections.

WHAT IS INCLUDED

● 14 sections, 185 lectures, and 20.5 hours of on-demand content covering the complete forecasting curriculum from fundamentals to 100,000-series hierarchical ML

● 57 downloadable resources: Excel workbooks, R project files, and real forecasting datasets for every section

● Graded assignments in every section — all on real time series data, with full solution walkthroughs

● Machine learning forecasting with Tidymodels — Section 13, added August 2023 by student demand, covering model stacking, cross-validation, and future prediction

● Fable package multi-model workflows: Prophet, VAR, hierarchical reconciliation, and 100,000-series automation in Sections 11, 12, and 14

● Lifetime access to all content and any future curriculum updates

● 30-day money-back guarantee — no questions asked

● Certificate of completion upon finishing the course

YOUR INSTRUCTOR

Haytham Omar, Ph.D.

Supply Chain & Business Intelligence Consultant · Developer · Trainer — UAE & France · Founder, Rescale Analytics

Haytham holds a Ph.D. in Supply Chain and Forecasting from the University of Bordeaux — his doctoral research was specifically in forecasting methodology and its application to supply chain planning. He also holds a Master of Science in Global Supply Chain Management from Bordeaux École de Management.

He is an active consultant whose forecasting models have been deployed by Sharaf Group Adventure HQ (replenishment algorithms since 2019), Sephora France (omni-channel demand planning, Ph.D. collaboration), and Aster Group. He has trained over 70,000 supply chain professionals across 70+ workshops in the UAE. Additional clients include DNO, Qarar, PWC Training Academy, and the Higher College of Technology.

He is also the creator of the Inventorize package for R and Python — used by over 90,000 supply chain professionals worldwide. This forecasting course is the R twin of his Python forecasting course, and together they form the most complete supply chain forecasting program available on any online learning platform.

Happy Forecasting — from Excel intuition to 100,000 time series in R.

14 sections · 20.5 hours · Excel + R + Fable + Tidymodels · Ph.D. instructor · Udemy for Business · 4.6 ★