How does it work – Complete Intelligence

How does it work?

The technology behind Complete Intelligence is built on three pillars:

A globally integrated machine learning model

A data-driven process without human intervention in the output

A simple means of interfacing with the platform

In general terms, how does the forecasting process work for RevenueFlow and CostFlow?

Receive historical financial data from an enterprise

Pre-Process

Data Validation
- Review for missing data
- Review for consistency of data
- Review for duplication of data
- Check formatting of data
Data Repair
- Crosswalk line items
- Interpolation of data
- Elimination of duplicates
- Correct formatting
Segregation of data
- Forecastable
- Not forecastable
- Forecastable after repair

Forecasting

Back-test individual algorithms within historical data of individual line item
Back-test combinations of individual algorithms within historical data of individual line item
Identify best combination of individual algorithms of back-test forecasts
Forecast future periods of individual line item based on best combination of individual algorithms identified
Repeat the above for each individual line item to be forecasted

Post-Process

Test forecast of individual line items for anomalies
- Outliers
- Positive and Negative
- Straight line / Identical slope
Run Forecast process again for individual line items identified

Issue Forecast Data

Forecast data is appended to historical data (same format and file type)