Future development suggestions

[Jsos17]

Future development suggestions

• Explore further ways to automatically update tutorials and other handwritten material when code base changes (For example, the code of some examples is automatically transferred to the user manual. Develop similar protocols.)

• Test and/or run examples automatically after code changes

• User can change the order of dataframe columns

• Explain dpEmu terminology in detail

• Add instructions for how to select suitable error parameter values

• Calculate sample statistics from data (sample mean, sample standard deviation etc) to guide selection of error parameter scale

• Smart and useful error messages for users

• Possibly remove setting of random generator seeds in the __init__ of abstract Filter class, if it does not have an adverse effect on reproducibility of results.

• Find the best way to visualize results

• Add support for real-time data

• dpEmu in the cloud

• 3D-plotting

• Add the option to use same random-generation seed in all applications of a filter under a series-node (At the end of tutorial II, procedures should be implementable without copying. In the current implementation there is also a problem with parallelism.)

• Document typical issues in data

• Consider to change the implementation of Filters, so that data is not modified in-place

• Add a Filter to simulate anonymization of data (for example, make GPS data more inaccurate)

• Integrate Google AutoML

• Add a function to find out when an alteration in error parameters causes a change in clustering

• Error generation as a command-line tool