As a general principle, the more time and effort a tasks takes, the more it should be weighted. Make sure the task is not unreasonably large as the students will have tasks in every subject to complete! 

A similar principle is that the more parts of the curriculum covered in a task, the more it might be weighted. 

Weight more discriminating and reliable tasks more heavily. 

Heavily weighted tasks need strong risk management procedures to ensure academic integrity. 

Some schools assemble many small tasks each with a small weighting into single portfolio task with a larger weighting. This has risks and benefits. A risk with many small tasks assembled into a portfolio or summative score, is that the small tasks will not sufficiently differentiate students. Another pitfall is  that they might not be of consistent and comparable standard and distort the picture of student capacity.  Designing a portfolio of tasks of sufficient challenge and consistent standard in a laborious task, therefore fewer parts are generally better than many parts.  To support students and prevent them doing it all at once at the end, it also requires regular grades based feedback to ensure students understand their progress.  Another concern with this model is that formative assessment becomes conflated with summative assessment thereby placing too much pressure on students who will not get a chance to learn through trial and error in a low stakes environment.  Feedback on progress is lost in the fog of high stakes assessment. Ultimately, There are more effective strategies for classroom management problems than making everything assessable.