Welcome to OMEN!
If you are already familiar with the system you can direct here http://omen-project.eu. Otherwise, please spare a minute to have a look at the following information.
OMEN is a web-based forecasting tool developed by FSU members to help experts, students and practitioners in the field of forecasting exploit basic and advanced forecasting techniques in everyday life in a direct and systematic way.
What’s the philosophy behind OMEN?
Forecasting support systems (FSSs) are used to assist forecasting procedures and support important decisions of companies. Despite their improvement through the years, vulnerabilities are still observed in three key aspects:
- Non web-based solutions
- Lack of customizability
- Complex user interfaces
As a forecasting unit we strongly believe that it is time to tackle these problems by exploiting and promoting modern solutions. In this respect we developed OMEN to indicate how by exploiting free open-source solutions a powerful FSS can be developed, offering a lot of advantages both in technological and methodological dimension. Thus, through OMEN we demonstrate the limitations of typical off-the-shelf FSSs and explain how some of its features could be used to deal with the abovementioned weaknesses and improve them.
More specifically, OMEN is a fully customizable web-based FSS with a modern and efficient interface which uses exclusively open-source solutions (developed under Shiny, R) and supports state-of-the-art methods and models. Our solution consist an adaptable and easy to customize web-based tool enabling 24/7 access of multiple users simultaneously by the device of their choice. A simplified interface organized in comprehensive and well documented menu items facilitate its use for practitioners, while state-of-the-art algorithms, such as temporal aggregation, intermittent demand and continuous time series forecasting models, become easily available using the cutting-edge functions provided by academics and researches to the community of R.
What’s the features of OMEN?
- Home: Create an account (or sign in if you are already a user) to upload time series of your interest, analyze them, forecast and retrieve the data and the results latter.
- Import Time Series: Import your personal data to the system and use OMEN to predict their future values.
- Statistics: Apply a high level statistical analysis to your data to inspect their primary characteristics.
- Transformations: In case of high variations, transformations can be used to rescale the original data, simplify their patterns and make them more consistent across the time series.
- Handling Outliers: Abnormalities in time series can spoil the patterns of the data and introduce a significant penalty to the forecasting performance. Detect them and get rid of them using the OMEN algorithms.
- Decompose: Decompose your time series and provide a clear picture of its main components (seasonal, trend, cyclical and randomness)
- Smoothing: Shrink the component of noise in your data to emphasize the useful characteristics of your time series (level and trend) and boost the total forecasting performance.
- Temporal aggregation: Transform the original data to alternative time frequencies, forecast the individual time series created and combine them to get your final predictions. Improvements in forecasting accuracy may rise even when simple forecasting models are used.
- Forecasting (Continuous Time Series): Decide which forecasting method will be used to predict your original-adjusted-normalized-deseasonalized-smoothed time series, according to the selections made so far in the previous tabs. Both traditional and state of the art methods can be found in the OMEN’s toolbox.
- Forecasting (Intermittent Demand Time Series): Are you dealing with intermittent demand data such as services and inventories which cannot be effectively handled by the continuous time series forecasting models? Try using some of the Intermittent Demand Time Series forecasting models included in the toolbox of OMEN.
- Judgmental adjustments: Do you want to take into consideration influential factors which are excluded from the statistical calculation of your forecasts? Do you think the forecasts produced so far are too optimistic or pessimistic? Try adjusting them.
- Error Metrics and Assessment: Inspect the performance of your forecasting method, its accuracy and bias using both numeric measurements and graphs.
- Report: Get a detailed report of your results (data, time series analysis, selected options and parameterization, forecasts and adjustments) and save it locally for latter use.