SEAMS Financials is a new product that combines forecasting of management accounts and the impact of finances on a company’s hard assets; it uses outputs from SEAMS WiLCO, such as optimal investment plans and forecasted asset key performance indicators (KPIs) for each modelled scenario. To report asset KPI trends SEAMS require the outputs from WiLCO to be approximated by functions of their independent variables so that clients may quickly explore the impacts of a range of investment scenarios.
SEAMS observed that their existing interpolation schemes would produce unreliable estimates in certain circumstances. They wanted an alternative method that, in addition to providing an accurate prediction, would give a measure of the uncertainty of that prediction.
The Smith Institute reviewed the current methods used by SEAMS for KPI fitting and error estimation, and then identified and evaluated the suitability of alternative methods to overcome short comings in the present methods. We selected the most suitable methods of fitting and error estimation, expressed them mathematically with respect to the asset KPI application, and then implemented them as algorithms in MATLAB. Finally, we evaluated the specification and performance of the implemented algorithms using examples of data from the SEAMS target application.
The methods reviewed and the tests on the least squares fitting algorithm and KPI models made it clear that the KPI for a given year is not only dependent upon the investment profile, but is also upon the KPI in the previous year.
As a result of this review, the Smith Institute provided SEAMS with potential models, together with examples of predictions and uncertainties. We also formalised the existing approach, allowing uncertainty estimates to be produced; and were able to demonstrate that the new models could achieve superior fits, even with fewer parameters. As a result of this work, SEAMS are able efficiently to produce predictions of KPI values, with higher accuracy and with a quantitative measure of the uncertainty associated with each prediction.
As a result of this work, SEAMS are able efficiently to produce predictions of KPI values, with higher accuracy and with a quantitative measure of the uncertainty associated with each prediction.
At SEAMS, we needed a rapid, accurate interpolation algorithm to enable asset service measure approximation in Financials. Smith Institute analysed SEAMS requirements quickly, researching the manifold options before selecting and proving the appropriate algorithm. Their solution gave our Financials key functionality, backed by professional and proven statistical skills. This reduced the time-to-market and improved the statistical robustness of our product. At the end of the day, Smith Institute did exactly what we asked them to – and they were quick and very thorough.Mark TurnerTechnology Director at SEAMS Ltd