Forecasting Financial Performance with Machine Learning for $10B Oil & Gas Organization

"Developed methodology to increase forecast accuracy and manage costs"

Client’s Financial Planning and Analysis (FP&A) team executed a monthly forecasting exercise largely dependent on qualitative inputs from sales and executive leadership
Leadership was concerned with the poor forecast accuracy and wanted to identify opportunities for improvement
The challenge for the forecasting team was identifying the operational factors driving revenue and costs across various data sets
FP&A leadership asked SLKone for process and systems improvement opportunities to increase forecast accuracy

Executed working sessions with key FP&A stakeholders to understand the process, systems, and data gaps
Established baseline forecast accuracy metrics based on historical performance
Aggregated operational data from across the business
Developed methodology leveraging random forest to identify natural split points which can more confidently predict revenue and cost targets
Trained a forecast model using vector auto regression based on the variables from the random forest exercise

90%+ predictive accuracy at three months for a significant cost line item based on an operational metric allowing the management team to better control costs before they were incurred
Defined limits and guard rails that would allow managers to view operational metrics and identify areas of concern for future financial performance

Case studies and expertise may represent projects and engagements performed by SLKone team members prior to their employment at SLKone