Awesome
TSFrames.jl
TSFrames is a Julia package to handle timeseries data. It provides a
convenient interface for the commonly used timeseries data
manipulations. TSFrames is built on top of the powerful and mature
DataFrames.jl making
use of the many capabilities of the DataFrame type and being easily
extensible at the same time.
Installing TSFrames
julia> using Pkg
julia> Pkg.add("TSFrames")
Basic usage
Creating TSFrame objects
TSFrames is a Tables.jl compatible package. This helps in easy conversion between TSFrame objects and other Tables.jl compatible types. For example, to load a CSV into a TSFrame object, we do the following.
julia> using CSV, Dates, DataFrames, TSFrames, Artifacts
julia> ts = CSV.read(joinpath(artifact"ibm", "IBM.csv"), TSFrame)
252x6 TSFrame with Date Index
Index Open High Low Close Adj Close Volume
Date Float64 Float64 Float64 Float64 Float64 Int64
─────────────────────────────────────────────────────────────────────
2021-04-26 136.157 137.314 135.258 135.344 129.028 4927497
2021-04-27 135.459 136.291 134.56 135.765 129.429 4062664
2021-04-28 136.635 137.094 135.851 136.711 130.332 3941433
2021-04-29 137.792 142.199 136.692 137.897 131.462 4554179
2021-04-30 137.38 137.505 134.369 135.641 129.311 9280321
2021-05-03 137.486 139.34 137.237 138.384 131.927 5997241
2021-05-04 138.059 140.143 137.983 139.34 132.838 6642623
2021-05-05 139.522 139.522 138.595 138.834 132.355 5229895
2021-05-06 138.872 141.979 138.795 141.893 135.272 7848661
2021-05-07 139.503 139.713 138.212 139.063 134.055 7325661
⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮
2022-04-11 127.95 128.18 126.18 126.37 126.37 3202500
2022-04-12 126.42 127.34 125.58 125.98 125.98 2691000
2022-04-13 125.64 126.67 124.91 126.14 126.14 3064900
2022-04-14 128.93 130.58 126.38 126.56 126.56 6382800
2022-04-18 126.6 127.39 125.53 126.17 126.17 4884200
2022-04-19 126.08 129.4 126.0 129.15 129.15 7971400
2022-04-20 135.0 139.56 133.38 138.32 138.32 17859200
2022-04-21 138.23 141.88 137.35 139.85 139.85 9922300
2022-04-22 139.7 140.44 137.35 138.25 138.25 6505500
233 rows omitted
As another example of this, consider the following code, which converts a TimeArray object to a TSFrame object. For this, we use the MarketData.yahoo function from the MarketData.jl package, which returns a TimeArray object.
julia> using TSFrames, MarketData;
julia> TSFrame(MarketData.yahoo(:AAPL); issorted = true)
10550×6 TSFrame with Date Index
Index Open High Low Close AdjClose Volume
Date Float64 Float64 Float64 Float64 Float64 Float64
───────────────────────────────────────────────────────────────────────────────────
1980-12-12 0.128348 0.128906 0.128348 0.128348 0.100039 4.69034e8
1980-12-15 0.12221 0.12221 0.121652 0.121652 0.09482 1.75885e8
1980-12-16 0.113281 0.113281 0.112723 0.112723 0.087861 1.05728e8
1980-12-17 0.115513 0.116071 0.115513 0.115513 0.090035 8.64416e7
1980-12-18 0.118862 0.11942 0.118862 0.118862 0.092646 7.34496e7
1980-12-19 0.126116 0.126674 0.126116 0.126116 0.0983 4.86304e7
1980-12-22 0.132254 0.132813 0.132254 0.132254 0.103084 3.73632e7
1980-12-23 0.137835 0.138393 0.137835 0.137835 0.107434 4.69504e7
1980-12-24 0.145089 0.145647 0.145089 0.145089 0.113088 4.80032e7
1980-12-26 0.158482 0.15904 0.158482 0.158482 0.123527 5.55744e7
1980-12-29 0.160714 0.161272 0.160714 0.160714 0.125267 9.31616e7
1980-12-30 0.157366 0.157366 0.156808 0.156808 0.122222 6.888e7
1980-12-31 0.152902 0.152902 0.152344 0.152344 0.118743 3.57504e7
1981-01-02 0.154018 0.155134 0.154018 0.154018 0.120048 2.16608e7
⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮
2022-09-27 152.74 154.72 149.95 151.76 151.76 8.44427e7
2022-09-28 147.64 150.64 144.84 149.84 149.84 1.46691e8
2022-09-29 146.1 146.72 140.68 142.48 142.48 1.28138e8
2022-09-30 141.28 143.1 138.0 138.2 138.2 1.24705e8
2022-10-03 138.21 143.07 137.69 142.45 142.45 1.14312e8
2022-10-04 145.03 146.22 144.26 146.1 146.1 8.78301e7
2022-10-05 144.07 147.38 143.01 146.4 146.4 7.9471e7
2022-10-06 145.81 147.54 145.22 145.43 145.43 6.84022e7
2022-10-07 142.54 143.1 139.45 140.09 140.09 8.58591e7
2022-10-10 140.42 141.89 138.57 140.42 140.42 7.4899e7
2022-10-11 139.9 141.35 138.22 138.98 138.98 7.70337e7
2022-10-12 139.13 140.36 138.16 138.34 138.34 7.04337e7
2022-10-13 134.99 143.59 134.37 142.99 142.99 1.13224e8
2022-10-14 144.31 144.52 138.19 138.38 138.38 8.85123e7
10522 rows omitted
Since we know that our data is in chronological order, we set the issorted keyword argument to the TSFrame constructor to true, allowing it to skip sorting the input table.
Indexing
julia> ts[1:10, [:Close]]
(10 x 1) TSFrame with Dates.Date Index
Index Close
Date Float64
─────────────────────
2021-04-26 135.344
2021-04-27 135.765
2021-04-28 136.711
2021-04-29 137.897
2021-04-30 135.641
2021-05-03 138.384
2021-05-04 139.34
2021-05-05 138.834
2021-05-06 141.893
2021-05-07 139.063
Subsetting
julia> from = Date(2021, 04, 29); to = Date(2021, 06, 02);
julia> TSFrames.subset(ts, from, to)
24x6 TSFrame with Date Index
Index Open High Low Close Adj Close Volume
Date Float64 Float64 Float64 Float64 Float64 Int64
────────────────────────────────────────────────────────────────────
2021-04-29 137.792 142.199 136.692 137.897 131.462 4554179
2021-04-30 137.38 137.505 134.369 135.641 129.311 9280321
2021-05-03 137.486 139.34 137.237 138.384 131.927 5997241
2021-05-04 138.059 140.143 137.983 139.34 132.838 6642623
2021-05-05 139.522 139.522 138.595 138.834 132.355 5229895
2021-05-06 138.872 141.979 138.795 141.893 135.272 7848661
2021-05-07 139.503 139.713 138.212 139.063 134.055 7325661
2021-05-10 139.388 141.855 139.388 139.742 134.709 7304636
2021-05-11 138.614 138.805 136.616 137.878 132.912 7454214
2021-05-12 137.514 137.811 134.933 135.086 130.221 6233742
2021-05-13 135.229 138.528 135.067 137.83 132.866 4807207
2021-05-14 138.728 139.283 137.629 138.317 133.336 2873780
2021-05-17 138.088 139.388 137.983 138.728 133.733 4471755
2021-05-18 138.413 138.91 136.931 137.581 132.627 4000009
2021-05-19 136.061 136.902 134.723 136.893 131.963 4498532
2021-05-20 136.826 138.537 135.908 137.553 132.599 4301675
2021-05-21 137.935 139.293 137.935 138.375 133.392 4219041
2021-05-24 138.681 138.996 137.839 138.356 133.373 3449290
2021-05-25 138.547 138.623 136.902 137.467 132.516 4118311
2021-05-26 137.189 137.658 136.75 137.075 132.138 3225655
2021-05-27 137.495 138.403 137.314 137.495 132.544 5889294
2021-05-28 137.868 137.983 137.18 137.419 132.47 2651192
2021-06-01 138.623 139.417 137.428 137.849 132.885 2528705
2021-06-02 138.26 139.34 137.772 139.312 134.295 2915097
Frequency conversion
julia> ts_weekly = to_weekly(ts)
52x6 TSFrame with Date Index
Index Open_last High_last Low_last Close_last Adj Close_last Volume_last
Date Float64 Float64 Float64 Float64 Float64 Int64
─────────────────────────────────────────────────────────────────────────────────────
2021-04-30 137.38 137.505 134.369 135.641 129.311 9280321
2021-05-07 139.503 139.713 138.212 139.063 134.055 7325661
2021-05-14 138.728 139.283 137.629 138.317 133.336 2873780
⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮
2022-04-14 128.93 130.58 126.38 126.56 126.56 6382800
2022-04-22 139.7 140.44 137.35 138.25 138.25 6505500
47 rows omitted
Plotting
julia> using Plots
julia> plot(ts_weekly[:, [:Close_last]], size = (600, 400))
Documentation
Head to the TSFrames user guide for more examples and functionality. The API reference is available on the documentation page.
Contributions
All or any contributions are welcome, small or large. Please feel free to fork the repository and submit a Pull Request.
JuliaCon
JuliaCon 2022 :
JuliaCon 2023 :
Acknowledgements
We gratefully acknowledge the JuliaLab at MIT for financial support for this project.
We thank Achim Zeileis, Jeffrey A. Ryan, Joshua M. Ulrich, Ross Bennett, and Corwin Joy for their remarkable work in the zoo and xts packages in R, which shaped our thinking in this field.
We thank Bogumił Kamiński and the Julia community for the remarkable DataFrames.jl package which was the foundation of our work, and for Prof. Kamiński's continuous peer review and feedback in our work.
We also thank all the code/documentation contributors to this package as well as the overall Julia community for all the valuable discussions.