SAP Analytics Cloud Predictive Scenarios democratises machine learning — letting business analysts build classification models (such as bank churn prediction) without deep technical expertise.
Machine learning and AI are increasingly popular as organisations seek innovative solutions. SAP Analytics Cloud (SAC) Predictive Scenarios democratises advanced analytics by removing technical barriers — enabling business analysts to generate meaningful insight through a user-friendly interface rather than complex algorithm development.
What are Predictive Scenarios?
The tool offers easy-to-implement classical ML methods with a user-friendly design, supporting three approaches: classifications, regressions and time-series forecasts. Users create and analyse data scenarios without manual sorting or extensive technical expertise.
Classification overview
Classification sorts data into distinct categories. In banking, this identifies customers likely to churn versus those who remain — analysing binary discrete variables such as an 'Exited' column, while calculating influencer contributions that reveal which factors most strongly affect outcomes.
Implementation example: bank churn
The example uses 7,999 customer records across 14 columns (7 measures, 7 dimensions). Exploratory analysis precedes training — outlier detection and statistical type classification (nominal, ordinal, continuous, textual). Model training then follows standard steps: create the classification model, add training data, select the predictive goal, and exclude unnecessary influencer columns.
Model evaluation
Predictive Power measures precision (values near 100% indicate confidence); Prediction Confidence indicates consistency on similar datasets; the Gini Index (0–1) suggests balanced class distribution near 0.5. The trained model then applies to test datasets of identical structure, generating prediction outputs to support decision-making.
Conclusion
This churn analysis demonstrates SAC Predictive Scenarios' ability to deliver actionable insight through machine learning — without requiring advanced technical qualifications.
