Skip to content

Latest commit

 

History

History
49 lines (33 loc) · 1.83 KB

README.md

File metadata and controls

49 lines (33 loc) · 1.83 KB
Raw

series

Useful Haskell functions for interacting with timed data.

A Series is a sequence of DataPoints, each DataPoint contains the time the data point occurs at, and the value. Reasoning about this structure, we can create various domain-specific operations to perform on these.

Features

  • Time slicing
  • Mapping
  • Resampling
  • Merging data
  • Importing/exporting from CSV
  • Diagram generation

Performance

The current implementation is backed by a Vector, and (ab)uses the O(1) slicing, and binary search in order to make operations more optimal than a naive implementation.

As an example, slicing Series with a list leads to O(n) slice. With a Vector, and binary search, we get O(log n) slice.

Here is a comparison of naive and vector-backed slicing:

benchmarking slice/series
time                 271.5 ns   (270.1 ns .. 273.2 ns)
                     1.000 R²   (1.000 R² .. 1.000 R²)
mean                 272.5 ns   (271.5 ns .. 273.6 ns)
std dev              3.600 ns   (2.779 ns .. 4.583 ns)
variance introduced by outliers: 13% (moderately inflated)

benchmarking slice/naive
time                 1.119 ms   (1.113 ms .. 1.126 ms)
                     0.999 R²   (0.999 R² .. 1.000 R²)
mean                 1.106 ms   (1.102 ms .. 1.114 ms)
std dev              21.08 μs   (16.63 μs .. 30.04 μs)

More benchmarks may follow.

Prior work

  • time-series (github private, no active maintainer)

    The primary inspiration for this package. The underlying data structure is a list.

  • timeseries (github)

    Seemingly very similar to time-series. Also uses a list. Has a notion of "time resolution" which is not yet implemented in this library.