shashi
commented
7 years ago Thanks for opening this issue!
I believe IndexedTables is a great fit for representing time-series data because of its sorted nature. In fact, time-series is a major use case considered in the design of IndexedTables thus far. JuliaTS.jl was an experiment by Tanmay in the early stages of this package worth revisiting now.
A time-series implementation could work on IndexedTables directly - an IndexedTable indexed by Date, DateTime, or Time columns (and possibly other columns of different types) would act as a time-series. This makes array-like and relational data operations in this package readily usable -- already very useful functionality.
The other half of the story is time-series operations such as those here -- I daresay most of them look straightforward to implement on top of IndexedTables. A next (optional) step is to wrap these operations in JuliaDB.jl which will give us a distributed time-series implementation.
Since you have expertise in this area, you are very welcome to contribute these operations if you're willing! We might get to doing this eventually, but presently this is not a pressing concern. A time-series roadmap issue might be in order.
Cheers
femtotrader
Hello,
https://github.com/tanmaykm/JuliaTS.jl implemented some time series and relational algebra operations over NDSparseData.
I wonder if there is some efforts in this direction.
Kind regards