Egides support you in your overall approach or on key stages in the construction of your project.
- The evaluation of your data: The quantity and the quality of your data are elements to be evaluated initially. They allow to estimate the level of performance that your tool can reach.
- Determination of a goal: The information to be predicted can be defined before the evaluation of the data (a data mining can then be organized to add information already available) or they can be chosen at the end of the inventory of data already available.
- Data analysis: The observations on which the model trains and then be tested, are analyzed a priori to evaluate their coherence and identify those can be considered as outliners. The relationships between the features of the observations are then studied, to make possible to identify the information having the most influence on the value to predict.
- Modeling: The construction of mathematical tools to answer this problem of optimization under stress, to reach the best compromise (quality - delay - cost - performance).
- Validation of the model: At the end of the training phase, the model is tested on non used sample for its training. The results make it possible to qualify the quality of the tool.
- Applying in production: The exploitation of the model in its environment (the one of the company) is achieved. User interfaces between several information systems of the company are monitored and enhanced as needed.
To explore the potential of these prediction tools, you can visit the application examples page.
