{"id":18612371,"url":"https://github.com/winvector/vtreat","last_synced_at":"2025-05-15T11:02:16.391Z","repository":{"id":19484080,"uuid":"22729841","full_name":"WinVector/vtreat","owner":"WinVector","description":"vtreat is a data frame processor/conditioner that prepares real-world data for predictive modeling in a statistically sound manner. Distributed under choice of GPL-2 or GPL-3 license.","archived":false,"fork":false,"pushed_at":"2025-01-09T22:21:41.000Z","size":181210,"stargazers_count":285,"open_issues_count":0,"forks_count":45,"subscribers_count":22,"default_branch":"main","last_synced_at":"2025-04-07T08:11:29.198Z","etag":null,"topics":["categorical-variables","machine-learning-algorithms","nested-models","prepare-data","r"],"latest_commit_sha":null,"homepage":"https://winvector.github.io/vtreat/","language":"HTML","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/WinVector.png","metadata":{"files":{"readme":"README.Rmd","changelog":"NEWS.md","contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2014-08-07T17:21:05.000Z","updated_at":"2025-02-01T18:54:01.000Z","dependencies_parsed_at":"2022-08-21T06:50:40.534Z","dependency_job_id":"a377cbb5-1e9d-4658-adfe-6c80d7746053","html_url":"https://github.com/WinVector/vtreat","commit_stats":{"total_commits":1055,"total_committers":6,"mean_commits":"175.83333333333334","dds":"0.049289099526066305","last_synced_commit":"7ee77dcef3082a9e791aa8f0d764d8910b12b326"},"previous_names":[],"tags_count":48,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WinVector%2Fvtreat","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WinVector%2Fvtreat/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WinVector%2Fvtreat/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WinVector%2Fvtreat/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/WinVector","download_url":"https://codeload.github.com/WinVector/vtreat/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":248910473,"owners_count":21181911,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["categorical-variables","machine-learning-algorithms","nested-models","prepare-data","r"],"created_at":"2024-11-07T03:16:50.218Z","updated_at":"2025-04-14T15:51:39.230Z","avatar_url":"https://github.com/WinVector.png","language":"HTML","readme":"---\noutput: github_document\n---\n\n\u003c!-- README.md is generated from README.Rmd. Please edit that file --\u003e\n\n\n\n```{r, echo = FALSE}\nknitr::opts_chunk$set(\n  collapse = TRUE,\n  comment = \" # \",\n  fig.path = \"README-\"\n)\n\n```\n\n[![DOI](https://joss.theoj.org/papers/10.21105/joss.00584/status.svg)](https://doi.org/10.21105/joss.00584)\n[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.1173313.svg)](https://doi.org/10.5281/zenodo.1173313)\n[![CRAN_Status_Badge](https://www.r-pkg.org/badges/version/vtreat)](https://cran.r-project.org/package=vtreat)\n[![status](https://tinyverse.netlify.com/badge/vtreat)](https://CRAN.R-project.org/package=vtreat)\n\n`vtreat` is a `data.frame` processor/conditioner (available [for `R`](https://github.com/WinVector/vtreat), and [for `Python`](https://github.com/WinVector/pyvtreat)) that prepares real-world data for supervised machine learning or predictive modeling in a statistically sound manner.\n\nA nice video lecture on what sorts of problems `vtreat` solves can be found [here](https://youtu.be/sniHkkrAsOc?t=42).\n\n`vtreat` takes an input `data.frame`\nthat has a specified column called \"the outcome variable\" (or \"y\")\nthat is the quantity to be predicted (and must not have missing\nvalues).  Other input columns are possible explanatory variables\n(typically numeric or categorical/string-valued, these columns may\nhave missing values) that the user later wants to use to predict \"y\".\nIn practice such an input `data.frame` may not be immediately suitable\nfor machine learning procedures that often expect only numeric\nexplanatory variables, and may not tolerate missing values.\n\nTo solve this, `vtreat` builds a transformed `data.frame` where all\nexplanatory variable columns have been transformed into a number of\nnumeric explanatory variable columns, without missing values.  The\n`vtreat` implementation produces derived numeric columns that capture\nmost of the information relating the explanatory columns to the\nspecified \"y\" or dependent/outcome column through a number of numeric\ntransforms (indicator variables, impact codes, prevalence codes, and\nmore).  This transformed `data.frame` is suitable for a wide range of\nsupervised learning methods from linear regression, through gradient\nboosted machines.\n\nThe idea is: you can take a `data.frame` of messy real world data and\neasily, faithfully, reliably, and repeatably prepare it for machine\nlearning using documented methods using `vtreat`.  Incorporating\n`vtreat` into your machine learning workflow lets you quickly work\nwith very diverse structured data.\n\nIn all cases (classification, regression, unsupervised, and multinomial classification) the intent is that `vtreat` transforms are essentially one liners.\n\nThe preparation commands are organized as follows:\n\n\n * **Regression**: [`R` regression example, fit/prepare interface](https://github.com/WinVector/vtreat/blob/master/Examples/Regression/Regression_FP.md), [`R` regression example, design/prepare/experiment interface](https://github.com/WinVector/vtreat/blob/master/Examples/Regression/Regression.md), [`Python` regression example](https://github.com/WinVector/pyvtreat/blob/master/Examples/Regression/Regression.md).\n * **Classification**: [`R` classification example, fit/prepare interface](https://github.com/WinVector/vtreat/blob/master/Examples/Classification/Classification_FP.md), [`R` classification example, design/prepare/experiment interface](https://github.com/WinVector/vtreat/blob/master/Examples/Classification/Classification.md), [`Python` classification  example](https://github.com/WinVector/pyvtreat/blob/master/Examples/Classification/Classification.md).\n * **Unsupervised tasks**: [`R` unsupervised example, fit/prepare interface](https://github.com/WinVector/vtreat/blob/master/Examples/Unsupervised/Unsupervised_FP.md), [`R` unsupervised example, design/prepare/experiment interface](https://github.com/WinVector/vtreat/blob/master/Examples/Unsupervised/Unsupervised.md), [`Python` unsupervised example](https://github.com/WinVector/pyvtreat/blob/master/Examples/Unsupervised/Unsupervised.md).\n * **Multinomial classification**: [`R` multinomial classification example, fit/prepare interface](https://github.com/WinVector/vtreat/blob/master/Examples/Multinomial/MultinomialExample_FP.md), [`R` multinomial classification example, design/prepare/experiment interface](https://github.com/WinVector/vtreat/blob/master/Examples/Multinomial/MultinomialExample.md), [`Python` multinomial classification example](https://github.com/WinVector/pyvtreat/blob/master/Examples/Multinomial/MultinomialExample.md).\n \nIn all cases: variable preparation is intended to be a \"one liner.\"\n\nThese current revisions of the examples are designed to be small, yet complete.  So as a set they have some overlap, but the user can rely mostly on a single example for a single task type.\n\n\nFor more detail please see here: [arXiv:1611.09477\nstat.AP](https://arxiv.org/abs/1611.09477) (the documentation describes the `R` version,\nhowever all of the examples can be found worked in `Python` [here](https://github.com/WinVector/pyvtreat/tree/master/Examples/vtreat_paper1)).\n\n`vtreat` is available\nas an [`R` package](https://github.com/WinVector/vtreat),\nand also as \na [`Python`/`Pandas` package](https://github.com/WinVector/vtreat).\n\n![](https://github.com/WinVector/vtreat/raw/master/tools/vtreat.png)\n\n(logo: Julie Mount, source: \"The Harvest\" by Boris Kustodiev 1914)\n\nEven with modern machine learning techniques (random forests, support vector\nmachines, neural nets, gradient boosted trees, and so on) or standard\nstatistical methods (regression, generalized regression, generalized\nadditive models) there are *common* data issues that can cause modeling\nto fail. vtreat deals with a number of these in a principled and\nautomated fashion.\n\nIn particular vtreat emphasizes a concept called\n\"y-aware pre-processing\" and implements:\n\n-   Treatment of missing values through safe replacement plus indicator\n    column (a simple but very powerful method when combined with\n    downstream machine learning algorithms).\n-   Treatment of novel levels (new values of categorical variable seen\n    during test or application, but not seen during training) through\n    sub-models (or impact/effects coding of pooled rare events).\n-   Explicit coding of categorical variable levels as new indicator\n    variables (with optional suppression of non-significant indicators).\n-   Treatment of categorical variables with very large numbers of levels\n    through sub-models (again [impact/effects\n    coding](https://win-vector.com/2012/07/23/modeling-trick-impact-coding-of-categorical-variables-with-many-levels/)).\n-   (optional) User specified significance pruning on levels coded into\n    effects/impact sub-models.\n-   Correct treatment of nested models or sub-models through data split\n    (see\n    [here](https://winvector.github.io/vtreat/articles/vtreatOverfit.html)) or\n    through the generation of \"cross validated\" data frames (see\n    [here](https://winvector.github.io/vtreat/articles/vtreatCrossFrames.html));\n    these are issues similar to what is required to build statistically\n    efficient stacked models or super-learners).\n-   Safe processing of \"wide data\" (data with very many variables, often\n    driving common machine learning algorithms to over-fit) through [out\n    of sample per-variable significance estimates and user controllable\n    pruning](https://winvector.github.io/vtreat/articles/vtreatSignificance.html)\n    (something we have lectured on previously\n    [here](https://github.com/WinVector/WinVector.github.io/tree/master/DS)\n    and\n    [here](https://win-vector.com/2014/02/01/bad-bayes-an-example-of-why-you-need-hold-out-testing/)).\n-   Collaring/Winsorizing of unexpected out of range numeric inputs.\n-   (optional) Conversion of all variables into effects (or \"y-scale\")\n    units (through the optional `scale` argument to `vtreat::prepare()`,\n    using some of the ideas discussed\n    [here](https://win-vector.com/2014/06/02/skimming-statistics-papers-for-the-ideas-instead-of-the-complete-procedures/)).\n    This allows correct/sensible application of principal component\n    analysis pre-processing in a machine learning context.\n-   Joining in additional training distribution data (which can be\n    useful in analysis, called \"catP\" and \"catD\").\n\nThe idea is: even with a sophisticated machine learning algorithm there\nare *many* ways messy real world data can defeat the modeling process,\nand vtreat helps with at least ten of them. We emphasize: these problems\nare already in your data, you simply build better and more reliable\nmodels if you attempt to mitigate them. Automated processing is no\nsubstitute for actually looking at the data, but vtreat supplies\nefficient, reliable, documented, and tested implementations of many of\nthe commonly needed transforms.\n\nTo help explain the methods we have\nprepared some documentation:\n\n-   The [vtreat package\n    overall](https://winvector.github.io/vtreat/index.html).\n-   [Preparing data for analysis using R white-paper](https://winvector.github.io/DataPrep/EN-CNTNT-Whitepaper-Data-Prep-Using-R.pdf)\n-   The [types of new\n    variables](https://winvector.github.io/vtreat/articles/vtreatVariableTypes.html)\n    introduced by vtreat processing (including how to limit down to\n    domain appropriate variable types).\n-   Statistically sound treatment of the nested modeling issue\n    introduced by any sort of pre-processing (such as vtreat itself):\n    [nested over-fit\n    issues](https://winvector.github.io/vtreat/articles/vtreatOverfit.html)\n    and a general [cross-frame\n    solution](https://winvector.github.io/vtreat/articles/vtreatCrossFrames.html).\n-   [Principled ways to pick significance based pruning\n    levels](https://winvector.github.io/vtreat/articles/vtreatSignificance.html).\n\nData treatments are \"y-aware\" (use distribution relations between\nindependent variables and the dependent variable).  For binary\nclassification use `designTreatmentsC()` and for numeric regression\nuse `designTreatmentsN()`.\n\nAfter the design step, `prepare()` should be used as you would use\nmodel.matrix.  `prepare()` treated variables are all numeric and never\ntake the value NA or +-Inf (so are very safe to use in modeling).\n\nIn application we suggest splitting your data into three sets: one for\nbuilding vtreat encodings, one for training models using these\nencodings, and one for test and model evaluation.\n\n\nThe purpose of `vtreat` library is to reliably prepare data for\nsupervised machine learning. We try to leave as much as possible to the\nmachine learning algorithms themselves, but cover most of the truly\nnecessary typically ignored precautions. The library is designed to\nproduce a `data.frame` that is entirely numeric and takes common\nprecautions to guard against the following real world data issues:\n\n-   Categorical variables with very many levels.\n\n    We re-encode such variables as a family of indicator or dummy\n    variables for common levels plus an additional [impact code](https://win-vector.com/2012/07/23/modeling-trick-impact-coding-of-categorical-variables-with-many-levels/)\n    (also called \"effects coded\"). This allows principled use\n    (including smoothing) of huge categorical variables (like zip-codes)\n    when building models. This is critical for some libraries (such as\n    `randomForest`, which has hard limits on the number of\n    allowed levels).\n\n-   Rare categorical levels.\n\n    Levels that do not occur often during training tend not to have\n    reliable effect estimates and contribute to over-fit.  vtreat\n    helps with 2 precautions in this case.  First the `rareLevel`\n    argument suppresses levels with this count our below from\n    modeling, except possibly through a grouped contribution.  Also\n    with enough data vtreat attempts to estimate out of sample\n    performance of derived variables.  Finally we suggest users\n    reserve a portion of data for vtreat design, separate from any\n    data used in additional training, calibration, or testing.\n\n-   Novel categorical levels.\n\n    A common problem in deploying a classifier to production is: new\n    levels (levels not seen during training) encountered during\n    model application. We deal with this by encoding categorical\n    variables in a possibly redundant manner: reserving a dummy variable\n    for all levels (not the more common all but a reference\n    level scheme). This is in fact the correct representation for\n    regularized modeling techniques and lets us code novel levels as all\n    dummies simultaneously zero (which is a reasonable thing to try).\n    This encoding while limited is cheaper than the fully Bayesian\n    solution of computing a weighted sum over previously seen levels\n    during model application.\n\n-   Missing/invalid values NA, NaN, +-Inf.\n\n    Variables with these issues are re-coded as two columns. The first\n    column is clean copy of the variable (with missing/invalid values\n    replaced with either zero or the grand mean, depending on the user\n    chose of the `scale` parameter). The second column is a dummy or\n    indicator that marks if the replacement has been performed. This is\n    simpler than imputation of missing values, and allows the downstream\n    model to attempt to use missingness as a useful signal (which it\n    often is in industrial data).\n\n-   Extreme values.\n\n    Variables can be restricted to stay in ranges seen during training.\n    This can defend against some run-away classifier issues during\n    model application.\n\n-   Constant and near-constant variables.\n\n    Variables that \"don't vary\" or \"nearly don't vary\" are suppressed.\n\n-   Need for estimated single-variable model effect sizes\n    and significances.\n\n    It is a dirty secret that even popular machine learning techniques\n    need some variable pruning (when exposed to very wide data frames,\n    see\n    [here](https://win-vector.com/2014/02/01/bad-bayes-an-example-of-why-you-need-hold-out-testing/)\n    and [here](https://www.youtube.com/watch?v=X_Rn3EOEjGE)). We make\n    the necessary effect size estimates and significances easily\n    available and supply initial variable pruning.\n\nThe above are all awful things that often lurk in real world data.\nAutomating these steps ensures they are easy enough that you actually\nperform them and leaves the analyst time to look for additional data\nissues. For example this allowed us to essentially automate a number of\nthe steps taught in chapters 4 and 6 of [*Practical Data Science with R*\n(Zumel, Mount; Manning 2014)](https://win-vector.com/practical-data-science-with-r/) into a\n[very short\nworksheet](https://winvector.github.io/KDD2009/KDD2009RF.html) (though we\nthink for understanding it is *essential* to work all the steps by hand\nas we did in the book). The 2nd edition of *Practical Data Science with R* covers\nusing `vtreat` in `R` in chapter 8 \"Advanced Data Preparation.\"\n\nThe idea is: `data.frame`s prepared with the\n`vtreat` library are somewhat safe to train on as some precaution has\nbeen taken against all of the above issues. Also of interest are the\n`vtreat` variable significances (help in initial variable pruning, a\nnecessity when there are a large number of columns) and\n`vtreat::prepare(scale=TRUE)` which re-encodes all variables into effect\nunits making them suitable for y-aware dimension reduction (variable\nclustering, or principal component analysis) and for geometry sensitive\nmachine learning techniques (k-means, knn, linear SVM, and more). You\nmay want to do more than the `vtreat` library does (such as Bayesian\nimputation, variable clustering, and more) but you certainly do not want\nto do less. \n\nThere have been a number of recent substantial improvements to the\nlibrary, including:\n\n-   Out of sample scoring.\n-   Ability to use `parallel`.\n-   More general calculation of effect sizes and significances.\n\n\nSome of our related articles (which should make clear some of our\nmotivations, and design decisions):\n\n-   [Modeling trick: impact coding of categorical variables with many levels](https://win-vector.com/2012/07/23/modeling-trick-impact-coding-of-categorical-variables-with-many-levels/)\n-   [A bit more on impact coding](https://win-vector.com/2012/08/02/a-bit-more-on-impact-coding/)\n-   [vtreat: designing a package for variable treatment](https://win-vector.com/2014/08/07/vtreat-designing-a-package-for-variable-treatment/)\n-   [A comment on preparing data for classifiers](https://win-vector.com/2014/12/04/a-comment-on-preparing-data-for-classifiers/)\n-   [Nina Zumel presenting on vtreat](https://www.slideshare.net/ChesterChen/vtreat)\n-   [What is new in the vtreat library?](https://win-vector.com/2015/05/07/what-is-new-in-the-vtreat-library/)\n-   [How do you know if your data has signal?](https://win-vector.com/2015/08/10/how-do-you-know-if-your-data-has-signal/)\n\nExamples of current best practice using `vtreat` (variable coding, train, test split) can be found [here](https://winvector.github.io/vtreat/articles/vtreatOverfit.html) and [here](https://winvector.github.io/KDD2009/KDD2009RF.html).\n\n\nSome small examples:\n\n\nWe attach our packages.\n\n```{r}\nlibrary(\"vtreat\")\npackageVersion(\"vtreat\")\ncitation('vtreat')\n```\n\nA small categorical example.\n\n```{r}\n# categorical example\nset.seed(23525)\n\n# we set up our raw training and application data\ndTrainC \u003c- data.frame(\n  x = c('a', 'a', 'a', 'b', 'b', NA, NA),\n  z = c(1, 2, 3, 4, NA, 6, NA),\n  y = c(FALSE, FALSE, TRUE, FALSE, TRUE, TRUE, TRUE))\n\ndTestC \u003c- data.frame(\n  x = c('a', 'b', 'c', NA), \n  z = c(10, 20, 30, NA))\n\n# we perform a vtreat cross frame experiment\n# and unpack the results into treatmentsC\n# and dTrainCTreated\nunpack[\n  treatmentsC = treatments,\n  dTrainCTreated = crossFrame\n  ] \u003c- mkCrossFrameCExperiment(\n  dframe = dTrainC,\n  varlist = setdiff(colnames(dTrainC), 'y'),\n  outcomename = 'y',\n  outcometarget = TRUE,\n  verbose = FALSE)\n\n# the treatments include a score frame relating new\n# derived variables to original columns\ntreatmentsC$scoreFrame[, c('origName', 'varName', 'code', 'rsq', 'sig', 'extraModelDegrees', 'recommended')] %.\u003e%\n  knitr::kable(.)\n\n# the treated frame is a \"cross frame\" which\n# is a transform of the training data built \n# as if the treatment were learned on a different\n# disjoint training set to avoid nested model\n# bias and over-fit.\ndTrainCTreated %.\u003e%\n  head(.) %.\u003e%\n  knitr::kable(.)\n\n# Any future application data is prepared with\n# the prepare method.\ndTestCTreated \u003c- prepare(treatmentsC, dTestC, pruneSig=NULL)\n\ndTestCTreated %.\u003e%\n  head(.) %.\u003e%\n  knitr::kable(.)\n```\n\nA small numeric example.\n\n```{r}\n# numeric example\nset.seed(23525)\n\n# we set up our raw training and application data\ndTrainN \u003c- data.frame(\n  x = c('a', 'a', 'a', 'a', 'b', 'b', NA, NA),\n  z = c(1, 2, 3, 4, 5, NA, 7, NA), \n  y = c(0, 0, 0, 1, 0, 1, 1, 1))\n\ndTestN \u003c- data.frame(\n  x = c('a', 'b', 'c', NA), \n  z = c(10, 20, 30, NA))\n\n# we perform a vtreat cross frame experiment\n# and unpack the results into treatmentsN\n# and dTrainNTreated\nunpack[\n  treatmentsN = treatments,\n  dTrainNTreated = crossFrame\n  ] \u003c- mkCrossFrameNExperiment(\n  dframe = dTrainN,\n  varlist = setdiff(colnames(dTrainN), 'y'),\n  outcomename = 'y',\n  verbose = FALSE)\n\n# the treatments include a score frame relating new\n# derived variables to original columns\ntreatmentsN$scoreFrame[, c('origName', 'varName', 'code', 'rsq', 'sig', 'extraModelDegrees')] %.\u003e%\n  knitr::kable(.)\n\n# the treated frame is a \"cross frame\" which\n# is a transform of the training data built \n# as if the treatment were learned on a different\n# disjoint training set to avoid nested model\n# bias and over-fit.\ndTrainNTreated %.\u003e%\n  head(.) %.\u003e%\n  knitr::kable(.)\n\n# Any future application data is prepared with\n# the prepare method.\ndTestNTreated \u003c- prepare(treatmentsN, dTestN, pruneSig=NULL)\n\ndTestNTreated %.\u003e%\n  head(.) %.\u003e%\n  knitr::kable(.)\n```\n\nRelated work:\n\n  * Cohen J, Cohen P (1983). Applied Multiple Regression/Correlation Analysis For The Behavioral Sciences. 2 edition. Lawrence Erlbaum Associates, Inc. ISBN 0-89859-268-2.\n  * [\"A preprocessing scheme for high-cardinality categorical attributes in classification and prediction problems\"](https://dl.acm.org/doi/10.1145/507533.507538) Daniele Micci-Barreca; ACM SIGKDD Explorations, Volume 3 Issue 1, July 2001 Pages 27-32.\n  * [\"Modeling Trick: Impact Coding of Categorical Variables with Many Levels\"](https://win-vector.com/2012/07/23/modeling-trick-impact-coding-of-categorical-variables-with-many-levels/) Nina Zumel; Win-Vector blog, 2012.\n  * \"Big Learning Made Easy – with Counts!\", Misha Bilenko, Cortana Intelligence and Machine Learning Blog, 2015.\n\n## Installation\n\n\nTo install, from inside `R` please run:\n\n```{r inst1, eval=FALSE}\ninstall.packages(\"vtreat\")\n```\n\n\n## Note\n\nNotes on controlling `vtreat`'s cross-validation plans can be found [here](https://github.com/WinVector/vtreat/blob/master/Examples/CustomizedCrossPlan/CustomizedCrossPlan.md).\n\nNote: `vtreat` is meant only for \"tame names\", that is: variables and column names that are also valid *simple* (without quotes) `R` variables names.\n","funding_links":[],"categories":[],"sub_categories":[],"project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fwinvector%2Fvtreat","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fwinvector%2Fvtreat","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fwinvector%2Fvtreat/lists"}