Forecasting & Time Series Analysis


Forecasting & Time Series Analysis

1. Introduction

Exegetic Analytics is a Data Science consultancy specialising in data acquisition and augmentation, data preparation, predictive analytics and machine learning. Our services are used by a range of industries from Education to Security, Food Delivery to Politics. Our consultants are based in Durban and Cape Town and we engage with clients all over the world. Our products and services are used by a multitude of industries including Aerospace, Education, Finance, Food and Transport.

Exegetic Analytics also offers training, with experienced and knowledgeable facilitators. Our courses focus on practical applications, working through examples and exercises based on real-world datasets.

All of our training packages include access to:

  • our online development environment and
  • detailed course material which participants will have continued access to even once the training has concluded.

For more information about what we do, you can refer to our website.

These are some of the companies who have benefitted from our trainning:

Take a look at our full list of courses to see what other training we have on offer.

Contact Us

If this proposal is of interest to you or you would like to hear more about what we do you can get in touch on or +27 (0)83 350 7699.

2. Course Description

A Time Series model allows you to make predictions about the future based on observations from the past. These models have important applications in science, industry and commerce.

Prediction is very difficult, especially if it’s about the future.
— Niels Bohr, Nobel Laureate (Physics)

Niels Bohr was right: predicting the future is a tough problem. Some things, like lottery numbers, are inherently unpredictable. Others, like air temperatures and rainfall, are reasonably predictable. Time series analysis makes it possible to assess whether or not predictions are possible and, if they are, build a model which can generate informed predictions for the future with realistic estimates of uncertainty.

The best qualification of a prophet is to have a good memory.
— George Savile

This course will provide you with an understanding of the theory behind time series models and the ability to build such models in R. By the end of the course you’ll be able to select the appropriate model for your data, train a model and start making predictions.


Duration 2 days
Why attend?
  1. Understand tools and methods of forecasting.
  2. Select appropriate forecasting techniques.
  3. Predict trends and estimate uncertainty.
  4. Use forecasting to improve decision-making and strategic planning.
  5. Evaluate and iteratively improve forecast accuracy.
Who should attend? Anybody interested in predicting the future will benefit from this course.

  • Analysts and strategic planners.
  • Managers (sales, marketing, product, business, financial, HR, inventory and supply chain).
  • Researchers.
Requirements We assume that participants have prior experience with R, ideally having completed both the the Introduction to R and Data Wrangling courses.

Return to our list of courses.

Course Outline

3. Course Outline

Day 1

  • Introduction
    • What is a Time Series?
    • Examples
  • Time Series Objects
    • Base Representation
      • Creating a ts object
      • Visualisation: plots and seasonal plots
      • Data quality, outliers and missing data
    • Other Representations
      • xts
      • zoo
  • Correlation
    • Covariance and correlation
    • Cross-Correlation
      • Correlation and causality
      • Significance
    • Autocorrelation (ACF)
    • Partial autocorrelation (PACF)
  • Basic Models
    • White Noise
    • Random Walk
    • Stationarity
  • Decomposition
    • Moving averages and smoothing
    • Additive and Multiplicative Decomposition
    • Trend
    • Seasonality
    • STL Decomposition
    • Decomposition and Forecasting
  • Model Principles
    • Accuracy and Precision
    • Training and Testing
    • Accuracy metrics
    • Rare Events and Black Swans
  • Exponential Smoothing
    • Simulation
    • Model estimation and forecasting
    • Adding trend and seasonality: Holt and Holt-Winters methods
  • Autoregressive (AR) Models
    • Simulation
    • PACF and autoregression
    • Model estimation and forecasting
  • Moving Average (MA) Models
    • Simulation
    • ACF and moving average
    • Model estimation and forecasting

Day 2

  • ARIMA Models
    • Simulation
    • Integrating for non-stationarity
    • Model estimation and forecasting
    • Automating with auto.arima()
  • Exogenous Variables
    • Examples
    • Adding Exogenous Variables to a model
    • ARMAX and ARIMAX models
    • Predicting with Exogenous Variables
  • Prophet
    • Non-linear and changing trends
    • Uncertainties and MCMC
    • Multiple seasonalities
    • Holidays
    • Non-uniform sampling and missing data
    • Exogenous Variables
  • Hierarchical Models
    • Bottom-up approach
    • Top-down approach
    • Middle-out approach
  • Anomaly Detection

Book now!

Training Philosophy

Our training emphasises practical skills. So, although you'll be learning concepts and theory, you'll see how everything is applied in the real world. We will work through examples and exercises based on real datasets.


All you'll need is a computer with a browser and a decent internet connection. We'll be using an online development environment. This means that you can focus on learning and not on solving technical problems.

Of course, we are happy to help you get your local environment set up too! You can start by following these instructions.


The training package includes access to
  • our online development environment and
  • detailed course material (slides and scripts).

Return to our list of courses.