Dynamic Planning: Predict Signals

Enablement of Predict Signals on the Master model in PCR application (Planning, Consolidation, Reporting) is a prerequisite. The flag to enable on the PCR side is "EnableIntegratedMachineLearning". Following this, the Predict Signals must be enabled in Dynamic Planning. The configuration flag in Dynamic Planning is "Enable Predict for Dap".

What is Predict Signals?

Predict Signals is a data analysis tool that identifies and flags potential anomalies in your data, allowing you to take corrective actions based on data-driven insights. These anomalies may originate from typos, mistakes, erroneous formulas, or intentional changes in planning assumptions. The tool provides a detailed context for each anomaly by utilizing an intuitive graph, enabling you to visualize the anomalies and their impact. Furthermore, it facilitates collaboration with colleagues and peers, promoting informed decision-making through shared insights.

How do Predict Signals work?

The AI Engine of Predict Signals utilizes machine learning and statistical/mathematical modeling to generate a predicted range representing the expected data ranges for specific intersections of dimensions. Any data values outside this predicted range are identified as Signals. These Signals are further categorized based on their level of deviation from the predicted normal range, including High risk, Medium risk, and Low risk classifications. This categorization helps prioritize and assess the severity of the anomalies detected by Predict Signals.

AI Engine of Predict Signals

Predict Signals incorporates a comprehensive machine learning pipeline that encompasses the entire data science cycle, spanning from data pre-processing to model deployment and management. The model training phase within this pipeline is equipped with a wide array of time-series algorithms, statistical methods, and classical machine-learning techniques. Advanced ensemble and stacking methods are employed to ensure consistent and precise results. To optimize the model training time, Predict Signals leverages multi-threading and parallel executions. 

Advantages

Predict Signals enables scenario data analysis, anomaly detection, and improved projection accuracy by providing contextual information about the identified signals. It offers a consolidated data analysis report, supports collaboration with colleagues and data owners, and facilitates corrective actions.

• Timely identification of Anomalies: Predict Signals empowers you to analyze scenario data and quickly detect anomalies caused by typos, formula mistakes, or input errors. With its timely anomaly identification, the tool helps you prevent costly mistakes and maintain the accuracy and integrity of your data.
• Efficient Planning: Predict Signals enhances work efficiency by facilitating error resolution and peer collaboration. It allows customization of the dashboard to meet your specific requirements, enabling you to focus on the necessary data for streamlined decision-making.
• Collaborative Decisioning: Predict Signals promotes collaborative decision-making by facilitating easy collaboration with colleagues when a signal is a result of a conscious change in planning. This enables you to work together efficiently and take appropriate actions based on shared insights.
• AI-Powered Actionable Insights: With its AI capabilities, Predict Signals transforms your data into actionable insights. It empowers businesses to make informed, data-driven decisions by providing valuable insights and analysis derived from the data.

Prerequisites

Contact the Planful support team for the following setup:

• Enablement of Predict Signals on the Master model in PCR application (Planning, Consolidation, Reporting) is a prerequisite.
• The flag to enable on the PCR side is "EnableIntegratedMachineLearning".
• Predict Signals will be enabled by default for DAP (Direct Access to PCR) models if it's enabled on the PCR (Planning, Consolidation, Reporting) side.

Enabling Predict Signals

You must contact the Planful Support team to enable the Predict Signals functionality in your application.