Jube is an open-source transaction monitoring platform that automates machine learning while emphasizing the importance of understanding foundational ML concepts to maintain quantitative competence and demystify fraud prevention systems. This practical training course focuses on implementing analytical techniques in real business environments, using real-world case studies cantered on transaction monitoring while also exploring other datasets like housing prices, credit risk, and fraud simulation.
Participants will develop skills in data analysis, predictive analytics (regression, decision trees, neural networks), and model optimization using R - the industry-standard, cost-effective statistical language - along with its powerful ecosystem of packages and RStudio’s intuitive interface for comprehensive advanced analytics without licensing costs. The course bridges theory and practice, equipping analysts to deploy advanced analytics across financial use cases while maintaining transparency in machine learning processes.
Learning Outcomes:
- Understand foundational ML & statistical techniques, including linear/logistic regression, decision trees, Bayesian classifiers, neural networks, feature engineering, and model optimization.
- Gain practical implementation skills in R, covering data manipulation, visualization (ggplot2), statistical analysis, and building/evaluating predictive models for fraud, credit risk, and numeric forecasting.
- Explore advanced topics like unsupervised learning (anomaly detection with Bayesian networks & SVMs), Monte Carlo simulation for risk modelling, and model deployment (APIs, reports, operational integration).
Training Plan:
The following is an abridged version of the full Advanced Analytics with R Training Plan.
The three-day training program covers advances analytics from foundational to advanced topics. On the first day, participants learn the basics of R, data structures, and linear regression, applying them to case studies like the Boston Housing Dataset and stock portfolio analysis. The second day focuses on classification and machine learning, including logistic regression for fraud detection, decision trees (C5.0) for credit risk, and Naive Bayes for handwriting recognition. The final day delves into advanced techniques, such as neural networks, anomaly detection ( Bayesian networks, SVMs), Monte Carlo simulations for fraud modelling, and deploying models via APIs (Plumber) and reports. The course provides a comprehensive, hands-on approach to predictive analytics.
Day 1: Foundations of Advanced Analytics
The first day of the workshop will establish core competencies in numeric prediction through applied linear statistical techniques using real market datasets. Participants will develop practical advanced analytics skills focused on foundational linear methods for numeric value forecasting, with an emphasis on hands-on implementation rather than theoretical discussion. The training will cover essential linear regression techniques and basic statistical modelling approaches, all directly applied to actual data scenarios to ensure immediate, tangible skills development in market data analysis and numeric prediction. This practical foundation will enable attendees to competently generate and interpret numeric forecasts while understanding the underlying statistical principles.
- Descriptive, Machine Learning, Predictive and Prescriptive Analytics Introduction
- Getting Started with R
- Data Structures
- Loading, Shaping and Merging Data
- Summary Statistics and Basic Plots in R
- Abstraction and Transformations (aka Feature Engineering)
- Linear Regression
Day 2: Classification & Machine Learning
The second day of the workshop advances into predictive analytics using machine learning, shifting focus from numeric prediction to classification techniques. While building on linear foundations with logistic regression, the day primarily explores nonlinear methods including decision trees and Bayesian classifiers. Participants will progress from data visualization with ggplot2 through increasingly sophisticated algorithms, applying each technique to real-world cases like fraud detection and credit risk analysis. The curriculum balances theoretical understanding with practical implementation, using R to develop models for handwriting recognition, fraud prevention optimization, and financial risk assessment - equipping attendees with both the technical skills and business context to deploy these methods effectively.
- Pretty Plots with ggplot2 and rapid, visual, dataset exploration
- Logistic Regression
- R Naive Bayesian Classifiers and Laplace Estimator
- Splits, Probability and Decision Trees
Day 3: Advanced Techniques & Deployment
- Neural Networks
- Unsupervised Learning with Bayesian Network Anomaly Detection
- Supervised and Unsupervised Learning with Bayesian Network Anomaly Detection
- Monte Carlo Model Search and Prescriptive Analytics
- Integration to the Operation (Outputting Reports and API’s)
Pricing
- 3 Days Core Training: EUR 2400
- 4 Days Core and Developer Training: EUR 3200
Training delivered onsite. Travel and expenses bill separately and in advance.