The client had very large data sets containing a wide range of parameters of a chemical production process. A number of questions about the process were answered based on the data. In order to provide useful insights the data was processed and analysed. All parameters were initially considered to identify relevant parameters and provide the most comprehensive interpretation of the data. Recommendations based on the investigation have led to improvements in the efficiency of the production process.
The specifics of a data set such as the complexity or requirement for processing prior to analysis, determine the time spend on a project.
estimated cost for a similar project: 5k
Based on machine logged data of a large factory several concepts were created for models which duplicate the production process. These models can be used for the tuning of the production process by providing an estimate of the influence of changes to the process. Similar models can also be used to predict failure along the production process, or monitor the process automatically by referencing the measured values to modeled values. An extensive concept exploration of varying models including machine learning methods can show the AI possibilities for a specific process with certain available data.
The accuracy of models depends on the process itself and the available data. These in turn influence the range of applicable methods for forecasting models. The desired functions by the client and the range of associated methods determines the size of the workload required to find the best solution.
estimated cost for a similar project: 10k
An iterative process takes place at a client where a trial is used to inform the next trial till the desired outcome is achieved. Data available from the trail and error process has been used to create a model which estimates the parameters required for the iterative process to converge significantly faster to the desired outcome. This model has been implemented in a web based solution to provide easy access. Further support of the model is provided by making sure the model is up to date to provide the most accurate estimations of parameters.
estimated cost for a similar project: 20k for development and 1k per month for maintenance