Introduction
Develop the skills to help your organisation make better risk/reward decisions.
What is Predictive Analytics?
Mathematical tools that analyse current or historical facts to predict future unknown events. Using Support Vector Machines, usually through an Excel front end, organisations with large data sets, such as customer information systems can classify and predict likely behaviours and outcomes.
How can we use them?
Z/Yen Group developed one of the first tools for Predictive Analytics more than 10 years ago. We have been working in multiple sectors which include:
Until very recently using Predictive Analytics needed proprietary software and serious data skills. That is changing fast – the software tools are often free, and can be used by data jockeys with just a little additional training.
How to get started
Based on this experience and success, Z/Yen have developed a set of training sessions to enhance understanding of predictive analytic methods for various organisations. There are two main types of courses:
Recognising that every organisation is different, the course modules can be mixed and matched, to tailor exactly what your organisation needs to get out of them. With more than 100 modules available, the courses can deliver exactly what you need – the table on below provides examples showing the scope and breadth of modules available.
Predictive Analytics Draft Content Overview – Examples
| Topic | Content includes |
|---|---|
| Predictive Analytics |
What is predictive analytics? Origins of Statistical Learning Theory Classical theory of statistics vs. statistical learning theory Theory of Ill-Posed Problems |
| Business |
Clear your mind Why use predictive analytics? 6 stage approach to problem solving Concerns and approaches to economic improvement, business efficiency, effectiveness and innovation Predictive analytics and economy improvement, business efficiency, effectiveness and innovation |
| Implementation | Application of SVMs to Marketing, Product Development Customer Retention and Service Improvement, focusing on case studies including customers most likely to leave/re-join, anomalous trades/subscription payments and credit application approvals |
| Support Vector Machine | Types of SVM models; SVMs for Classification, Prediction and Anomaly detection; SVM constraints; Making predictions with SVM models; Evaluating Model performance; What are false positives and false negatives?; False positives vs. False negatives; SVM Platforms |
| R |
Installing R platform; Basic R commands; Data frames in R Data import and export in R; Building SVM models in R's kernlab; Making SVM predictions in R's kernlab; Troubleshooting and error messages in R |
| Puzzle | Bertrand's box paradox; Birthday Problem; Monty Hall; Simpson's Paradox; St. Petersburg Paradox – what are they and how do they relate to business concerns |
| Data | How to deal with data gaps; how to partition data effectively, Obtaining sufficient data; Selecting appropriate variables for prediction; Eliminating confirmation bias; Big data and sampling |
| Linear Algebra | What is a vector space?; What are matrices?; What is a dimension?; Linear transformations |
| Statistics | Bias and Skew; Interpreting Statistical measures; What is Correlation?; What is R-squared?; What is causation?; Correlation and causation |
| Probability | What is probability?; What is a distribution?; How do you use probabilities in simulations; Monte Carlo; Bayesian inference |
| Visualisation |
Scatter plots; Matrix plots; Histograms; Box plots; 3D data-maps; Profile and Cluster screens; Voronoi diagrams. |