Predictive Analytics with Orange is a course in building predictive models with Orange Data Mining by connecting widgets, from data preparation, regression and classification evaluated with cross-validation and explained with SHAP to clustering and working alongside Python. It suits data analysts and data scientists with basic Python who want results they can see and explain.
Predictive analytics helps organisations forecast sales, spot customers who are likely to leave or estimate risk from the data they already have. Orange Data Mining is an open source tool that builds this kind of work by dragging widgets onto a canvas and connecting them, with interactive results at every step. That makes it quick to grasp the idea behind each algorithm, and it can be extended with Python when a model needs to go into real use.
This course takes learners through the current Orange 3.40, from exploring and preparing data to building and comparing regression and classification models with cross-validation, ROC and confusion matrices, and explaining models with SHAP. It continues with neural networks, clustering, dimensionality reduction and association rules, then connects Orange to Python so that saved models can score new data. It closes with an end-to-end forecasting project on realistic data. (3 days, 6 hours per day, 18 hours in total, Beginner level.)
Course Details
This course is the 2026 edition of Predictive Analytics with Orange, updated for Orange 3.40. It runs for 3 days, 6 hours per day (18 hours in total, 09:00-16:00), as lectures with a lab in every topic. Beginner level, for learners with basic Python. Day one focuses on data preparation and regression. Day two covers classification, model evaluation, explaining models with SHAP and neural networks. Day three covers clustering, dimensionality reduction, association rules and using Orange with Python, with a capstone that scores new data. All tools are open source and free. Learners take home a lab guide, Orange workflows, Python code and sample datasets for every lab.
Section 1: Lab: Getting Started with Orange Data Mining
Section 2: Lab: Loading and Exploring Data
Section 3: Preparing Data for Modelling
Section 4: Lab: Building Regression Models
Section 5: Reading Models and Explaining Drivers
Section 6: Lab: Classification Algorithms
Section 7: Evaluating Classification Models
Section 8: Imbalanced Data and Explaining Models
Section 9: Lab: Neural Networks
Section 10: Workshop: Predicting Customer Churn
Section 11: Lab: Clustering
Section 12: Dimensionality Reduction
Section 13: Lab: Association Rules
Section 14: Lab: Using Orange with Python
Section 15: Workshop: Capstone Project
No public rounds are open right now. Join the waiting list and we will contact you first when the next round opens, or ask us on LINE. Or call 02-570-8449 or 088-807-9770
Supervised learning trains a model on data that already has the answer (the target), such as customer records where you know who churned, so it can predict the answer for new data. It covers regression and classification. Unsupervised learning has no answer to learn from; the model finds structure on its own, for example grouping customers with clustering or finding products bought together with association rules. The course covers both in Orange.
Load the data with File and set the target column with Select Columns, then prepare it with Preprocess. Connect learners such as Logistic Regression, Random Forest or Gradient Boosting to Test and Score to evaluate them with cross-validation. Check the errors in Confusion Matrix and ROC Analysis, then use Predictions to score new data and explain the results with SHAP from the Explain add-on.
k-Means needs the number of clusters up front and assigns each point to the nearest centre. It is fast and suits large datasets, and you can choose the number of clusters with the silhouette score. Hierarchical clustering merges the closest points step by step into a tree (a dendrogram), so you see structure at several levels and choose where to cut afterwards, but it is slower on large data. The course compares both with DBSCAN on the same customer data.
It suits data analysts, data scientists, business analysts, developers and teachers who want to build predictive models with a visual tool. You should be able to write basic Python, because the last day uses Orange with Python, and be comfortable with Excel and tables of data. Basic statistics will help you learn faster. Please bring a laptop on which you can install software.
You will be able to prepare data, build and compare regression and classification models, evaluate them with cross-validation, confusion matrices and ROC, explain the drivers with feature importance and SHAP, segment customers with clustering, find products bought together with association rules, and save models to score new data from Python. You also take home workflows and sample code for every lab.
Yes. IT Genius Institute runs in-house training for organisations. Lab problems and datasets can be adapted to your team's work, such as sales forecasting, churn prediction or customer segmentation, and the balance between Orange and Python can be adjusted to the learners' background. Contact the institute to request a quotation.
Python is one of the most widely used programming languages today, in web development, automation, data work, IoT and AI. Its readable syntax and gentle learning curve make it the first language of choice for many schools and organisations. Yet many beginners…
Most organisations hold large amounts of sales, customer and system usage data, yet often use it only for backward-looking reports. The same data can drive predictive models, for example forecasting sales, predicting which customers are likely to cancel,…
Machine learning is still at the heart of AI work in organisations, from sales forecasting and customer screening to document categorisation and anomaly detection, and a great deal of tabular data work is still handled better and more cheaply by classic…
