CausalTables.jl - v1.3.4

CausalTables v1.3.4

Diff since v1.3.3

Changes

• Simplify some syntax to avoid compilation issues with earlier Julia versions.
• Update NetworkSummary test to avoid dependence on random seed whose behavior may change with the version of Julia.

Merged pull requests: - minor syntax changes to ensure compatibility with Julia 1.6 (#54) (@salbalkus)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] 6 months ago

CausalTables.jl - v1.3.3

CausalTables v1.3.3

Diff since v1.3.2

Changes

• treat_all, treat_none, and all functions that rely on them (i.e. ate) now correctly keep treatment as a Bool instead of casting to Float64

Merged pull requests: - treat_all now assumes treatment is Bool instead of Float (#53) (@salbalkus)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] 7 months ago

CausalTables.jl - v1.3.2

CausalTables v1.3.2

Diff since v1.3.1

Changes

• Fix issue where summaries without specific targets (e.g. Friends) would not be added to causes when summarizing a CausalTable

Merged pull requests: - Summaries without targets bugfix (#52) (@salbalkus)

Closed issues: - Friends incorrectly filtered (#51)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] 8 months ago

CausalTables.jl - v1.3.1

CausalTables v1.3.1

Diff since v1.3.0

Changes and Bug Fixes

• select and functions that depend on it now retain summaries whose targets are selected
• Summaries are no longer duplicated in the causes attribute of a CausalTable
• ape now correctly implements samples parameter
• rand can be called on a DataGeneratingProcess without a sample size (will use $n = 1$ by default)

Merged pull requests: - Release 1.3.1 (#50) (@salbalkus)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] 8 months ago

CausalTables.jl - v1.3.0

CausalTables v1.3.0

Diff since v1.2.7

New features

• A step that draws from a $d$-dimensional MultivariateDistribution in a DataGeneratingProcess now creates $d$ columns, and the package now supports splitting a matrix generated by a distribution into multiple columns of a CausalTable.
• The symbol ≈ can be used to denote a step that is random but not drawn from a distribution (e.g. generating a random graph), so that its value in a CausalTable is not updated after intervention.
• String instead of Symbol now supported in construction of CausalTable and StructuralCausalModel, as well as calling condensity and related functions.
• One can now sample from a DataGeneratingProcess directly using the rand function, without needing to construct an SCM.

Bug fixes and other changes

• The arrays attribute of a CausalTable is now updated after applying an intervention.

Merged pull requests: - Release 1.3.0 (#49) (@salbalkus)

Closed issues: - Allow treatment, response, controls, etc. to be specified as strings instead of Symbols (#17) - Allow link functions (#19) - Intermediate variables not recomputed (#48)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] 8 months ago

CausalTables.jl - v1.2.7

CausalTables v1.2.7

Diff since v1.2.6

Changes

• Update ape and draw_counterfactual to work correctly when a summary function is applied to treatment (interventions are now propagated down the SCM).
• Correct typos within inline documentation
• Remove extraneous Project.toml file

Merged pull requests: - Release 1.2.7 (#47) (@salbalkus)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] 9 months ago

CausalTables.jl - v1.2.6

CausalTables v1.2.6

Diff since v1.2.5

New features

• select now also filters variables described in the summaries attribute of a CausalTable as well as the data.

Bug fixes and other changes

• Fix bug where confounders failed to collapse the confounders of a treatment and its summary to a single CausalTable object.
• Fix bug where summarize would add summarized variables to the causes attribute even if they were not causes originally (the function now includes an option to enable this behavior, disabled by default).
• Improved error description in condensity.

Merged pull requests: - Release 1.2.6 (#46) (@salbalkus)

Closed issues: - collapse_parents parameter not working for summarized treatments (#45)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] 9 months ago

CausalTables.jl - v1.2.5

CausalTables v1.2.5

Diff since v1.2.4

Changes

• Implement quality assurance testing with Aqua.jl
• Fix unnecessary exports
• Fix type piracy from a specialized version of the convolve function from Distributions.jl being defined for convolutions over vectors of distributions

Merged pull requests: - Minor changes to improve package quality, including tests with Aqua.jl (#43) (@salbalkus) - add [extra] for test dependencies in main Project.toml (#44) (@salbalkus)

Closed issues: - make DGPs parameterized? (#42)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] 10 months ago

CausalTables.jl - v1.2.4

CausalTables v1.2.4

Diff since v1.2.3

Changes

• Fixed bug that made it possible for the condensity of a given step in a DataGeneratingProcess to be computed based on variables that came after it in the sequence.
• Replaced TableTransforms dependency with NamedTupleTools.

Closed issues: - Bug when creating directly from O (#41)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] 11 months ago

CausalTables.jl - v1.2.3

CausalTables v1.2.3

Diff since v1.2.2

Changes

• Change how efficiency bounds are computed in the ate, att, and atu functions.
• Remove efficiency bound output where erroneous estimates were previously included (cfmean, cfdiff, and ape).
• Improve ground truth ate estimate to be more precise.

Merged pull requests: - Release 1.2.3 (#40) (@salbalkus)

Closed issues: - Inaccuracies in efficiency bound approximation (#39)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] 12 months ago

CausalTables.jl - v1.2.2

CausalTables v1.2.2

Diff since v1.2.1

New Features

• Fixed typos and tagline in documentation, README, and paper
• Preparation for JOSS

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] about 1 year ago

CausalTables.jl - v1.2.1

CausalTables v1.2.1

Diff since v1.2.0

New Features

• Changed replace to Base.replace to avoid errors when both are used in other packages.

Merged pull requests: - Release 1.2.1 (#38) (@salbalkus)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] about 1 year ago

CausalTables.jl - v1.2.0

CausalTables v1.2.0

Diff since v1.1.1

New Features

• CausalTables are now constructed with a causes parameter instead of confounders. This permits treatment-response pairs with different sets of confounders and supports the extraction of causal-relevant variables such as confounders, mediators, and instruments algorithmically based on the listed causes of each treatment and response. This will break existing code that specified confounders for a CausalTable using the confounders parameter.
• Added mediators and instruments along with corresponding "name" and "matrix" versions to extract mediators and instrumental variables from a given CausalTable.
• Added convenience constructors to make it easier to create DataGeneratingProcess objects with default names and labels.
• Added merge function to combine two DataGeneratingProcess objects into one.
• Can now call rand(scm) without specifying sample size for settings where the SCM is "data-adaptive" in the sense that parameters other than the first distribution within the DGP depend on the sample size.

Merged pull requests: - Release 1.2.0 (#37) (@salbalkus)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] about 1 year ago

CausalTables.jl - v1.1.1

CausalTables v1.1.1

Diff since v1.1.0

New Features

• Added parents function, which subsets the columns of a CausalTable to only those labeled as causes of a given variable.
• Updated paper.

Merged pull requests: - Release 1.1.1 (#36) (@salbalkus)

Closed issues: - Documentation Review (#20)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] about 1 year ago

CausalTables.jl - v1.1.0

CausalTables v1.1.0

Diff since v1.0.0

New Features

• Added intervene function to intervene on the treatment of an existing CausalTable. Currently implemented interventions include treat_all and treat_none (for binary treatments) as well as additive_mtp(δ) and multiplicative_mtp(δ) for continuous treatments. New interventions coming soon!
• Added propensity function to compute (generalized) propensity scores for a given variable in a CausalTable.
• Added responsematrix, treatmentmatrix, and confoundersmatrix functions for getting the response, treatment, and confounder variables in matrix form.
• Updated paper and documentation.

Merged pull requests: - Release 1.1.0 (#35) (@salbalkus)

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] over 1 year ago

CausalTables.jl - v1.0.0

CausalTables v1.0.0

The first stable release of CausalTables.jl. Implements the CausalTable, DataGeneratingProcess, and StructuralCausalModel data structures, along with associated methods. These support easy wrapping and simulating data with labeled causal structure.

Scientific Software - Peer-reviewed - Julia
Published by github-actions[bot] over 1 year ago