TDWI Certificate Track

Prerequisite: None

Course Outline

Accurate forecasts are the Holy Grail of data analytics. Why?

Because every organization uses forecasts to plan its activities:

• Short-term forecasts are used for staff scheduling and customer service.
• Medium-term forecasts are used for purchasing supplies and materials.
• Long-term forecasts are used for strategic decision-making.

If your team is serious about making an impact using data, you can’t go wrong learning how to craft robust forecasts.

Unfortunately, there’s a problem.

Traditional forecasting techniques often cannot handle the complexities of modern organizations. This is why organizations that need state-of-the-art forecasting are increasingly turning to machine learning.

This three-day workshop teaches your team the skills needed to craft state-of-the-art machine learning forecasting models using Python.

The first two days of the workshop provide the foundation, covering how ML algorithms work, how to engineer features for the best predictive machine learning models, and how to tune ML models for optimal predictive performance.

With this machine learning foundation in place, the workshop’s third day focuses on using machine learning models for forecasting. Topics include what makes ML forecasting different, simple forecasting methods, evaluating models, and engineering time series features.

Attendees develop practical, actionable skills through 12 hands-on labs that utilize free, open source software.

While this workshop incorporates some mathematics, the mathematical level is accessible to a broad audience, focusing on concepts rather than calculations.

Your Team Will Learn

• The different types of machine learning
• The two forms of supervised learning – classification and regression
• The CART classification tree algorithm
• The mathematics of classification trees
• Overfitting – the bugbear of machine learning
• The bias-variance tradeoff
• Tuning CART classification tree models
• Measuring the accuracy of your classification tree models
• Engineering predictive features for your decision tree models
• The CART regression tree algorithm
• The mathematics of regression trees
• The random forest algorithm
• Tuning random forest models
• What is time series forecasting?
• The factors that impact forecasts.
• How to evaluate your forecasting.
• Simple forecasting methods with Python.
• Why simple methods are usually not enough.
• Why machine learning is the future of forecasting.
• How to use machine learning models for forecasting.
• Resources to continue your learning.

Geared To

• Business and data analysts
• BI and analytics developers and managers
• Business users
• Aspiring data scientists
• Anyone interested in crafting useful forecasts

No background in advanced mathematics or statistics is required.

Prerequisites

This course requires basic knowledge of Python and Jupyter Notebooks, which can be acquired by completing a complimentary Python Quick Start online tutorial from TDWI.

Laptop Setup

Attendees must have a laptop computer with the required software installed prior to the session. In advance of the class, attendees will receive detailed instructions on downloading and installing the software.