We have developed this research environment for ourselves, and would like to share the productivity boost we have experienced with other researchers. In our experience the most time consuming parts of the process have been
- processing all the raw data files into an analysable format
- creating a test bench to look at results from all angles
These tasks can and should be automated, no reinventing the wheel for every new Masters student.
While these are rules of thumb, it may be helpful to have an indication of the time required for different steps
- Set-up is perhaps the trickiest part, which can take 2 hours and needs to be done carefully, especially the set-up of version control with Git. Setting up all devices to be ready to measure should take no more than 15 minutes.
- Collecting data, once one has done it before should take no more than 15 minutes to set up and explain to subjects, the actual recording takes as long as it takes, and then after the recording it can take another 15 minutes to save all the file recordings in a folder. If data streaming from the Everion hasn-t completed, it may take a bit longer until that download has happened
- Processing automatically should take perhaps a few seconds per recorded minute of data and is fully automated, except for the manual step of reviewing the annotations made. Our experience is that this should not take more than a couple of minutes per file.
- Analyse interactively can take up to 10 minutes to load all files into memory in the beginning (if there are only a handful of files it can be a few seconds), but should then allow an interactive exploration of all results with publication-ready charts.
- Develop code a generic functions is available, which has access to all sensor data and signals generated within the Everion at run-time. Coding just takes as long as it takes, but review is rapidly possible through interactive analysis. In addition, there is the option of optimising hyperparameters, where run/times depend on the number of iterations and the number of files to optimise over.
- Deploy code is a service which we offer, contingent on whether the codes is compilable under the specified circumstances. If code is compilable, results should be available within a few working days, if code has been developed using our development environment and without extra toolboxes.