best way to store large time series data

Question

Srinivas 2013-1-16

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/58918-best-way-to-store-large-time-series-data

I plan to collect intraday data (1 min) for several instruments. what is the best way to keep them for long term ?

should I keep appending in a mat file , with one mat file for each instrument or should I keep them as CSV in different text file for each instrument.

what would be more robust ?

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Jing 2013-1-16

What do you mean 'robust'? CSV is more compatible with other software, like Excel, and can open with plain text editor while mat-file can't. But mat-file can save all kinds of variables in MATLAB.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Pedro Villena 2013-1-16

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/58918-best-way-to-store-large-time-series-data#answer_71299

编辑：Pedro Villena 2013-1-16

The EAS3 file format is used to store floating point data in IEEE format and to exchange the files between different computer architectures (little/big endian). The data is organized as parameters with one parameter being a one-, two- or three-dimensional floating point array. Several of these parameters may be combined to one time step. This allows to store five-dimensional arrays. Data can be written in single-precision (32 Bit), double-precision (64 Bit) or quadruple-precision (128 Bit). Geometry information for the different directions are saved in the header of the file. It is also possible to store additional information in user defined arrays there. With the file size being limited only by the computer itself (e.g. file system), EAS3 files are suitable for large simulations and thus for high-performance computing.