Predictive Analysis

Harnessing the power of statistics to illuminate the unknown.

Predictive models are found in all walks of life these days, whether it’s models that calculate consumers’ likelihood to churn, models for recommending relevant digital content, or models for predicting what the weather will be like tomorrow. All of these applications share a common theme – using data about known characteristics to predict some unknown feature. Such predictions may involve forecasting future values or events but may equally be focussed on predicting unknown entities in the present.

This one-day course is designed to help research and insight professionals who interact with analytics to gain greater understanding of predictive models, from the background and terminology used in the field, to the fundamentals of how the models work and how they can be implemented in the commonly-used, open-source, statistical programming language R. As such, the course will briefly touch on the R language, although no prior knowledge is assumed.

Who will find this course useful:

This course will benefit client and agency side researchers and insight professionals with a quantitative and analytical focus.

Objectives:

• Understand different types of predictive modelling problems.
• Develop familiarity with a range of predictive algorithms, the theory behind them, and their implementation.
• Learn how to evaluate algorithm performance.

Learning outcomes:

The course will give delegates a theoretical overview and guidelines for applying a number of techniques from statistics and machine learning, including:

• Regression: linear, binomial, multinomial.
• Tree-based methods: decision trees, random forests, XGBoost.
• Neural networks: feedforward, convolutional, and recurrent networks.