Artelys Crystal Forecast
In a world that is contantly evolving, forecasters are increasingly requested to bring insights into the future and enlight business contexts. The growing volume of available data and the diversity of data sources is a real opportunity and, at the same time, a technical and a structural challenge for companies. Given this context, an agile use of computer science, statistical modeling and business expertise give forecasters a real opportunity to overcome those difficulties to produce analyses and forecasts with an unprecedented level of precision and relevance.
Relying on our knowledge of data analytics and statistical modeling, our powerful, innovative and adaptive technologies embeded in Artelys Crystal, and on your expertise of your Business, Artelys Crystal Forecast allows you to generate trustworthy forecasts for the short, mid and long-term horizons, covering both your operational and strategic planning needs. Thanks to a user-friendly interface and a guided process, it is easy to automatically exploit data flows, generate forecasts and gain visibility into your future! See more benefits
POWER - Improve forecasts accuracy and generate value
The efficiency of models embedded in Artelys Crystal Forecast, in the form of error minimization indicators and its adaptability, has been proven through the numerous studies carried out, particularly in the Energy sector for a diversity of actors, ranging from distributors to utilities.
The default models used are multiple non-parametric models, or multivariate regression models but other models, such as Unobserved Component Models, ARIMAX or neural networks, are available in our library.
Moreover, Artelys Crystal Forecast furnishes the popular statistical software R to provide you with the latest and most robust algorithms for model calibration and forecasting, and to enable you to create your own modeling scripts.
SIMPLICITY - Complex data management and analysis become finally easy
Artelys Crystal Forecast provides you with the highest level of guidance during the modelling process thanks to its organization in workflows. The tool connects to your environment, allowing to import data automatically, from your own databases but also from external sources, and a direct export of results.
Artelys experts have gathered an important library of built-in tools and models to manipulate your data easily accessible through our user-friendly interface. Your specific data cleansing/treatments or statistical models can also be easily integrated into the library in order to reproduce and automate the usual repetitive and tiresome processes that precede the modelling phase.
ADAPTABILITY - Configurable and scalable to your needs
The software interface and workflow are fully configurable so that variables, statistical indicators, visualization tools, models and forecasting parameters can be set to satisfy your modeling needs.
Artelys Crystal Forecast allows you to implement also a relational structure within your data and create an ad-hoc segmentation in order to depict accurately either your client portfolio, or the fundamental structure of your data.
Our Solution is built in such a way that it can deal with small datasets and aggregated data as well as with huge datasets at a very low granularity keeping its performance at the highest.
Easy integration to your own processes and IT
Artelys Crystal Forecast integrates a powerful workflow manager allowing to define and automate the different steps of the forecasting process: data management, model calibration, forecasting and visualization. Each step of the workflow is configurable, allowing you to reflect the realities of your business in the forecasting process.
Moreover, the Artelys Crystal Database manager allows to recover, correct, reconcile, store and organize the complete set of historical data, exogeneous variables and forecasts. It centralizes clean data from multiple and heterogenous sources, and provides a set of database connectors and files importer.
If needed, Artelys Crystal Server can be used for a Client-Server architecture, in order to automate the various workflows and move heavy computing on its server to address the upcoming needs of processing large amount of data.
Artelys Crystal Forecast can be connected to web interfaces to distribute forecast results to final users.
Artelys Crystal Forecast becomes the cockpit of the forecaster, where he can manage all the forecasting steps from data processing to forecast diffusion, and find all the tools he needs to perform deep analyses of historical data to extract information essential to forecasts.
Time series forecasting expertise and accuracy
Artelys Crystal Forecast has been developed by time series forecasting experts that were willing to build up their experience and their modeling skills into a unique powerful software:
• Well proven methodologies and models on real business context and data
• Strong accuracy of forecasts on test samples
• Script library available and editable to configure
Data cleansing, correction and transformation
Statistical modelling and optimization algorithm
• Model configurations with the help of the statistical model chooser screen
Fundamental structure in your data and best suited segmentation
Artelys Crystal Forecast allows you to take into account the realities of your business by letting you construct relational structures between your time series of interest:
Spatial segmentations. Your business territory is probably large, mixed and diverse. That’s why forecasts can be generated at any geographical from the very local to regional, national or transnational. The structuration of the data enables relational links between all your variables and makes it possible to reconcile at every aggregation level.
Client portfolio segmentation, allowing you to integrate your own business segmentation and potentially complete the segmention with quantitative clustering.
Typology/Usage/Equipment segmentation. This structure enables you to take into account the most uncertain cases in a long term view which is the modification of the fundamental structure of your energetical data.
Calendar, seasonal and weather effects can be integrated in the database and indexed with your own data, allowing you to take into account the intricacies of your business as well some of the sources of highest level of uncertainty in your forecasting process.
Configurable built-in statistical indicators and visualization tools to take a critical view on your forecast
Artelys Crystal Forecast is embedded in the Artelys Crystal visualization Suite, that ease your understanding of your data and the reporting of final results:
Entirely configurable visualization tools with built-in indicators to quantify, rank and visualize interactions between the main effects. Screen configurations include numerous graph views including time series, two-dimensional visualizations, as well as configurable correlation and regression graphs that can be embedded in the time series view.
Numerous aggregation and zoom options allowing you to get a perfect insight into your data from as many angles as possible. Data aggregations can be performed with any time step, including hourly, but also daily, weekly and monthly profiles.
A full range of statistical indicators to validate the forecast accuracy, including a summary table displaying the global quality of fitted parameters and confidence in the future forecasts.
Adaptive help messages to escort you in updates of the model.
Self-awareness of the statistical quality of your model with powerful indicators. These include statistical indicators (MAPE, RMSE), prediction power, error evaluation and model diagnostics indicators.
Automatic export of configurable PDF or Excel Reporting Dashboards
Improve your risk management process by reducing short-term forecast error
Erroneous demand forecasts can lead to important imbalances and high operation costs. Artelys Crystal Forecast helps you carry out more accurate forecasts and, doing so, allows you to reduce, both efficiently and profitably, your market operations costs.
Artelys Crystal Forecast is of particular interest for any utility or market participant that buys or sells power, and participate in the balancing mechanism. Its flexibility makes it applicable for intraday, day ahead or forward contracts positioning.
Even in non-market contexts, anticipation of possible tightness in the production-demand equilibrium, due to structural evolutions ou conjunctural events, is a matter of great interest in a day-to-day management, since its can lead to the use of expensive last-minute production assets or even loss of load.
Artelys Crystal Forecast helps you anticipate a possible disruptancy on energy supply due to the combinaison of simultaneous effect, and participate to the activation of consumption cut-off or a temporal re-organization of the network.
Represent mid-term uncertainty on demand evolution
While managing hydro-reservoir, fuel storage capacities, or take-or-pay contracts, an average forecast of demand evolution is not enouth to define mid-term management strategies ; a realistic modelization of uncertainty structure is necessary. Artelys Crystal Forecast enables you to generate a set of demand evolution scenarios, taking into account multiple uncertainty sources, such as tendencial growth hypothesis or weather conditions.
For instance, Artelys Crystal Forecast helps you generate hundreds of hourly power demand scenarios for next year, taking into account underlying temperature scenarios and economic growth hypothesis. It then enables you design and test your energy system robustness to mid-term uncertainty.
Create long-term demand scenarii for investment planning
Projection of energy demand under significant structural changes in the energetic/technologic/usage mix or macro evolution linked to economic growth or energetic efficiency is a great matter for 10-20 year strategic planning. Artelys Crystal Forecast enables you to depict a bottom-up structure and parametrize the evolution scenarios.
Artelys Crystal Forecast models helps key national and/or international actors to depict and simulate various scenarii of demand at long term horizon as an input to a capacity expansion planning optimization or to strategic study.
Manage your client portfolio variations through time
Minimize risk of oversizing/undersizing of ressources or means of supply anticipating portfolio growth or structural evolution, and their consequences on global demand. Artelys Crystal Forecast is applicable, even in non-energetic contexts:
Artelys Crystal Forecast models are used by a telecom company willing to order the right amount of ressources in the call centers every month taking into account the diversity/seasonality of the requests.
For instance, Artelys Crystal Forecast fits retailer needs to get an accurate forecast of sales by product or group of product for several type of clients in order to improve financial reporting and help its stores or warehouses to minimize stocks