{ "_id": "6a27d48124555f66ed545b98", "Package": "cramR", "Type": "Package", "Title": "Cram Method for Efficient Simultaneous Learning and Evaluation", "Version": "0.1.0", "Date": "2025-05-11", "Authors@R": "c(\nperson(\"Yanis\", \"Vandecasteele\", email = \"yanisvdc.ensae@gmail.com\", role = c(\"cre\", \"aut\")),\nperson(\"Michael Lingzhi\", \"Li\", email = \"mili@hbs.edu\", role = \"ctb\"),\nperson(\"Kosuke\", \"Imai\", email = \"imai@harvard.edu\", role = \"ctb\"),\nperson(\"Zeyang\", \"Jia\", email = \"zeyangjia@fas.harvard.edu\", role = \"ctb\"),\nperson(\"Longlin\", \"Wang\", email = \"longlin_wang@g.harvard.edu\", role = \"ctb\")\n)", "Maintainer": "Yanis Vandecasteele ", "Description": "Performs the Cram method, a general and efficient approach\nto simultaneous learning and evaluation using a generic machine\nlearning algorithm. In a single pass of batched data, the\nproposed method repeatedly trains a machine learning algorithm\nand tests its empirical performance. Because it utilizes the\nentire sample for both learning and evaluation, cramming is\nsignificantly more data-efficient than sample-splitting. Unlike\ncross-validation, Cram evaluates the final learned model\ndirectly, providing sharper inference aligned with real-world\ndeployment. The method naturally applies to both policy\nlearning and contextual bandits, where decisions are based on\nindividual features to maximize outcomes. The package includes\ncram_policy() for learning and evaluating individualized binary\ntreatment rules, cram_ml() to train and assess the\npopulation-level performance of machine learning models, and\ncram_bandit() for on-policy evaluation of contextual bandit\nalgorithms. For all three functions, the package provides\nestimates of the average outcome that would result if the model\nwere deployed, along with standard errors and confidence\nintervals for these estimates. Details of the method are\ndescribed in Jia, Imai, and Li (2024)\n\nand Jia et al. (2025) .", "License": "GPL-3", "URL": "https://github.com/yanisvdc/cramR,\nhttps://yanisvdc.github.io/cramR/", "BugReports": "https://github.com/yanisvdc/cramR/issues", "Encoding": "UTF-8", "RoxygenNote": "7.3.2", "VignetteBuilder": "knitr, rmarkdown", "Config/testthat/edition": "3", "Config/Needs/citation": "yes", "Config/pak/sysreqs": "cmake make libicu-dev libpng-dev libuv1-dev python3", "Repository": "https://yanisvdc.r-universe.dev", "Date/Publication": "2025-05-15 02:51:19 UTC", "RemoteUrl": "https://github.com/yanisvdc/cramr", "RemoteRef": "HEAD", "RemoteSha": "b6d7444fd9f0ea7eabec603642b62818caf8e724", "NeedsCompilation": "no", "Packaged": { "Date": "2026-06-09 08:47:32 UTC", "User": "root" }, "Author": "Yanis Vandecasteele [cre, aut],\nMichael Lingzhi Li [ctb],\nKosuke Imai [ctb],\nZeyang Jia [ctb],\nLonglin Wang [ctb]", "MD5sum": "1734bbd17570799e8a0f939c189ba8d0", "_user": "yanisvdc", "_type": "src", "_file": "cramR_0.1.0.tar.gz", "_fileid": "cad96955358a1749ed27c47218b0d5c822b85264d44420c8d391b0105f30b3c9", "_filesize": 3606500, "_sha256": "cad96955358a1749ed27c47218b0d5c822b85264d44420c8d391b0105f30b3c9", "_created": "2026-06-09T08:47:32.000Z", "_published": "2026-06-09T08:53:21.822Z", "_distro": "noble", "_jobs": [ { "job": 80283531690, "time": 248, "config": "linux-devel-x86_64", "r": "4.7.0", "check": "OK", "artifact": "7502914612" }, { "job": 80283531737, "time": 291, "config": "linux-release-x86_64", "r": "4.6.0", "check": "OK", "artifact": "7502931058" }, { "job": 80283531705, "time": 300, "config": "macos-oldrel-arm64", "r": "4.5.3", "check": "OK", "artifact": "7502919981" }, { "job": 80283531706, "time": 216, "config": "macos-release-arm64", "r": "4.6.0", "check": "OK", "artifact": "7502889675" }, { "job": 80282374676, "time": 399, "config": "source", "r": "4.6.0", "check": "OK", "artifact": "7502820606" }, { "job": 80283531644, "time": 175, "config": "wasm-release", "r": "4.6.0", "check": "OK", "artifact": "7502887095" }, { "job": 80283531729, "time": 227, "config": "windows-devel", "r": "4.7.0", "check": "OK", "artifact": "7502907641" }, { "job": 80283531712, "time": 165, "config": "windows-oldrel", "r": "4.5.3", "check": "OK", "artifact": "7502884238" }, { "job": 80283531698, "time": 189, "config": "windows-release", "r": "4.6.0", "check": "OK", "artifact": "7502892629" } ], "_buildurl": "https://github.com/r-universe/yanisvdc/actions/runs/27194410887", "_status": "success", "_host": "GitHub-Actions", "_upstream": "https://github.com/yanisvdc/cramr", "_commit": { "id": "b6d7444fd9f0ea7eabec603642b62818caf8e724", "author": "yanis vandecasteele ", "committer": "yanis vandecasteele ", "message": "up website with CRAN install\n", "time": 1747277479 }, "_maintainer": { "name": "Yanis Vandecasteele", "email": "yanisvdc.ensae@gmail.com", "login": "yanisvdc", "linkedin": "in/yanis-vandecasteele", "description": "Harvard MS in Data Science student\nWorking on differential privacy, ML and statistics (notably bayesian statistics and causal inference)", "uuid": 112790082 }, "_registered": true, "_dependencies": [ { "package": "R", "version": ">= 3.5.0", "role": "Depends" }, { "package": "caret", "version": ">= 7.0-1", "role": "Imports" }, { "package": "grf", "version": ">= 2.4.0", "role": "Imports" }, { "package": "glmnet", "version": ">= 4.1.8", "role": "Imports" }, { "package": "stats", "version": ">= 4.3.3", "role": "Imports" }, { "package": "magrittr", "version": ">= 2.0.3", "role": "Imports" }, { "package": "doParallel", "version": ">= 1.0.17", "role": "Imports" }, { "package": "foreach", "version": ">= 1.5.2", "role": "Imports" }, { "package": "DT", "version": ">= 0.33", "role": "Imports" }, { "package": "data.table", "version": ">= 1.16.4", "role": "Imports" }, { "package": "keras", "version": ">= 2.15.0", "role": "Imports" }, { "package": "dplyr", "version": ">= 1.1.4", "role": "Imports" }, { "package": "purrr", "role": "Imports" }, { "package": "R6", "role": "Imports" }, { "package": "rjson", "role": "Imports" }, { "package": "R.devices", "role": "Imports" }, { "package": "itertools", "role": "Imports" }, { "package": "iterators", "role": "Imports" }, { "package": "testthat", "version": ">= 3.0.0", "role": "Suggests" }, { "package": "covr", "version": ">= 3.5.1", "role": "Suggests" }, { "package": "kableExtra", "version": ">= 1.4.0", "role": "Suggests" }, { "package": "profvis", "version": ">= 0.4.0", "role": "Suggests" }, { "package": "devtools", "role": "Suggests" }, { "package": "waldo", "role": "Suggests" }, { "package": "knitr", "role": "Suggests" }, { "package": "rmarkdown", "role": "Suggests" }, { "package": "randomForest", "role": "Suggests" }, { "package": "gbm", "role": "Suggests" }, { "package": "nnet", "role": "Suggests" }, { "package": "withr", "role": "Suggests" } ], "_owner": "yanisvdc", "_selfowned": true, "_usedby": 0, "_updates": [], "_tags": [], "_stars": 3, "_contributors": [ { "user": "yanisvdc", "count": 312, "uuid": 112790082 } ], "_userbio": { "uuid": 112790082, "type": "user", "name": "Yanis Vandecasteele", "description": "Harvard MS in Data Science student\r\nWorking on differential privacy, ML and statistics (notably bayesian statistics and causal inference)" }, "_downloads": { "count": 192, "source": "https://cranlogs.r-pkg.org/downloads/total/last-month/cramR" }, "_devurl": "https://github.com/yanisvdc/cramr", "_pkgdown": "https://yanisvdc.github.io/cramR/", "_searchresults": 30, "_rbuild": "4.6.0", "_assets": [ "extra/citation.cff", "extra/citation.html", "extra/citation.json", "extra/citation.txt", "extra/contents.json", "extra/cramR.html", "extra/NEWS.html", "extra/NEWS.txt", "extra/readme.html", "extra/readme.md", "manual.pdf" ], "_homeurl": "https://github.com/yanisvdc/cramr", "_realowner": "yanisvdc", "_cranurl": true, "_releases": [ { "version": "0.1.0", "date": "2025-05-14" }, { "version": "0.1.1", "date": "2025-08-24" } ], "_exports": [ "BatchContextualEpsilonGreedyPolicy", "BatchContextualLinTSPolicy", "BatchLinUCBDisjointPolicyEpsilon", "ContextualLinearBandit", "cram_bandit", "cram_bandit_est", "cram_bandit_sim", "cram_bandit_var", "cram_estimator", "cram_expected_loss", "cram_learning", "cram_ml", "cram_policy", "cram_policy_value_estimator", "cram_simulation", "cram_var_expected_loss", "cram_variance_estimator", "cram_variance_estimator_policy_value", "fit_model", "fit_model_ml", "get_betas", "LinUCBDisjointPolicyEpsilon", "ml_learning", "model_predict", "model_predict_ml", "set_model", "test_baseline_policy", "test_batch", "validate_params", "validate_params_fnn" ], "_help": [ { "page": "BatchContextualEpsilonGreedyPolicy", "title": "Batch Contextual Epsilon-Greedy Policy", "topics": [ "BatchContextualEpsilonGreedyPolicy" ] }, { "page": "BatchContextualLinTSPolicy", "title": "Batch Contextual Thompson Sampling Policy", "topics": [ "BatchContextualLinTSPolicy" ] }, { "page": "BatchLinUCBDisjointPolicyEpsilon", "title": "Batch Disjoint LinUCB Policy with Epsilon-Greedy", "topics": [ "BatchLinUCBDisjointPolicyEpsilon" ] }, { "page": "ContextualLinearBandit", "title": "Contextual Linear Bandit Environment", "topics": [ "ContextualLinearBandit" ] }, { "page": "cram_bandit", "title": "Cram Bandit: On-policy Statistical Evaluation in Contextual Bandits", "topics": [ "cram_bandit" ] }, { "page": "cram_bandit_est", "title": "Cram Bandit Policy Value Estimate", "topics": [ "cram_bandit_est" ] }, { "page": "cram_bandit_sim", "title": "Cram Bandit Simulation", "topics": [ "cram_bandit_sim" ] }, { "page": "cram_bandit_var", "title": "Cram Bandit Variance of the Policy Value Estimate", "topics": [ "cram_bandit_var" ] }, { "page": "cram_estimator", "title": "Cram Policy Estimator for Policy Value Difference (Delta)", "topics": [ "cram_estimator" ] }, { "page": "cram_expected_loss", "title": "Cram ML Expected Loss Estimate", "topics": [ "cram_expected_loss" ] }, { "page": "cram_learning", "title": "Cram Policy Learning", "topics": [ "cram_learning" ] }, { "page": "cram_ml", "title": "Cram ML: Simultaneous Machine Learning and Evaluation", "topics": [ "cram_ml" ] }, { "page": "cram_policy", "title": "Cram Policy: Efficient Simultaneous Policy Learning and Evaluation", "topics": [ "cram_policy" ] }, { "page": "cram_policy_value_estimator", "title": "Cram Policy: Estimator for Policy Value (Psi)", "topics": [ "cram_policy_value_estimator" ] }, { "page": "cram_simulation", "title": "Cram Policy Simulation", "topics": [ "cram_simulation" ] }, { "page": "cram_var_expected_loss", "title": "Cram ML: Variance Estimate of the crammed expected loss estimate", "topics": [ "cram_var_expected_loss" ] }, { "page": "cram_variance_estimator", "title": "Cram Policy: Variance Estimate of the crammed Policy Value Difference (Delta)", "topics": [ "cram_variance_estimator" ] }, { "page": "cram_variance_estimator_policy_value", "title": "Cram Policy: Variance Estimate of the crammed Policy Value estimate (Psi)", "topics": [ "cram_variance_estimator_policy_value" ] }, { "page": "fit_model", "title": "Cram Policy: Fit Model", "topics": [ "fit_model" ] }, { "page": "fit_model_ml", "title": "Cram ML: Fit Model ML", "topics": [ "fit_model_ml" ] }, { "page": "get_betas", "title": "Generate Reward Parameters for Simulated Linear Bandits", "topics": [ "get_betas" ] }, { "page": "LinUCBDisjointPolicyEpsilon", "title": "LinUCB Disjoint Policy with Epsilon-Greedy Exploration", "topics": [ "LinUCBDisjointPolicyEpsilon" ] }, { "page": "ml_learning", "title": "Cram ML: Generalized ML Learning", "topics": [ "ml_learning" ] }, { "page": "model_predict", "title": "Cram Policy: Predict with the Specified Model", "topics": [ "model_predict" ] }, { "page": "model_predict_ml", "title": "Cram ML: Predict with the Specified Model", "topics": [ "model_predict_ml" ] }, { "page": "set_model", "title": "Cram Policy: Set Model", "topics": [ "set_model" ] }, { "page": "test_baseline_policy", "title": "Validate or Set the Baseline Policy", "topics": [ "test_baseline_policy" ] }, { "page": "test_batch", "title": "Validate or Generate Batch Assignments", "topics": [ "test_batch" ] }, { "page": "validate_params", "title": "Cram Policy: Validate User-Provided Parameters for a Model", "topics": [ "validate_params" ] }, { "page": "validate_params_fnn", "title": "Cram Policy: Validate Parameters for Feedforward Neural Networks (FNNs)", "topics": [ "validate_params_fnn" ] } ], "_readme": "https://github.com/yanisvdc/cramr/raw/HEAD/README.md", "_rundeps": [ "backports", "base64enc", "bslib", "cachem", "caret", "class", "cli", "clock", "codetools", "config", "cpp11", "crosstalk", "data.table", "diagram", "DiceKriging", "digest", "doParallel", "dplyr", "DT", "e1071", "evaluate", "farver", "fastmap", "fontawesome", "foreach", "fs", "future", "future.apply", "generics", "ggplot2", "glmnet", "globals", "glue", "gower", "grf", "gtable", "hardhat", "here", "highr", "htmltools", "htmlwidgets", "ipred", "isoband", "iterators", "itertools", "jquerylib", "jsonlite", "keras", "KernSmooth", "knitr", "labeling", "later", "lattice", "lava", "lazyeval", "lifecycle", "listenv", "lmtest", "lubridate", "magrittr", "MASS", "Matrix", "memoise", "mime", "ModelMetrics", "nlme", "nnet", "numDeriv", "otel", "parallelly", "pillar", "pkgconfig", "plyr", "png", "pROC", "processx", "prodlim", "progressr", "promises", "proxy", "ps", "purrr", "R.devices", "R.methodsS3", "R.oo", "R.utils", "R6", "rappdirs", "RColorBrewer", "Rcpp", "RcppEigen", "RcppTOML", "recipes", "reshape2", "reticulate", "rjson", "rlang", "rmarkdown", "rpart", "rprojroot", "rstudioapi", "S7", "sandwich", "sass", "scales", "shape", "sparsevctrs", "SQUAREM", "stringi", "stringr", "survival", "tensorflow", "tfautograph", "tfruns", "tibble", "tidyr", "tidyselect", "timechange", "timeDate", "tinytex", "tzdb", "utf8", "vctrs", "viridisLite", "whisker", "withr", "xfun", "yaml", "zeallot", "zoo" ], "_vignettes": [ { "source": "cram_bandit.Rmd", "filename": "cram_bandit.html", "title": "Cram Bandit", "engine": "knitr::rmarkdown", "headings": [ "What is cram_bandit()?", "Introduction: Bandits, Policies, and Cram", "Understanding the inputs", "Example: Use cram_bandit() on simulated data with batch size of 1", "References" ], "created": "2025-03-27 06:23:48", "modified": "2025-04-29 14:52:57", "commits": 5 }, { "source": "cram_bandit_helpers.Rmd", "filename": "cram_bandit_helpers.html", "title": "Cram Bandit Helpers", "engine": "knitr::rmarkdown", "headings": [ "🌟 What is this article about?", "πŸ› οΈPolicy parameters explained", "✨ Epsilon-Greedy Policy", "πŸ€– Theoretical computation", "πŸ“Š Code helper", "πŸ”’ LinUCB Disjoint Policy with ( \\varepsilon )-Greedy", "πŸ€“ Thompson Sampling (LinTS)", "πŸ‘¨β€πŸ’» Practical Workflow", "πŸ§ͺ Estimand Calculation in cram_bandit_sim()", "πŸ“‚ Useful Links", "🌟 Acknowledgments" ], "created": "2025-03-27 06:23:48", "modified": "2025-05-05 19:14:36", "commits": 8 }, { "source": "cram_bandit_simulation.Rmd", "filename": "cram_bandit_simulation.html", "title": "Cram Bandit Simulation", "engine": "knitr::rmarkdown", "headings": [ "What is cram_bandit_sim()?", "πŸ“‹ Inputs", "Example: Cram Bandit Simulation", "What Does It Return?", "Example Output Preview", "Notes", "Recommended Use Cases", "References" ], "created": "2025-03-27 06:23:48", "modified": "2025-05-11 18:54:22", "commits": 12 }, { "source": "cram_ml.Rmd", "filename": "cram_ml.html", "title": "Cram ML", "engine": "knitr::rmarkdown", "headings": [ "Cram ML", "Output of Cram ML", "Built-in Model", "loss_name argument", "Regression Losses", "Classification Losses", "Custom Loss Functions", "caret_params argument", "Case of categorical target variable", "Case 1: Predicting Class Labels", "Case 2: Predicting Class Probabilities", "Custom Model", "1. custom_fit(data, ...)", "2. custom_predict(model, data)", "3. custom_loss(predictions, data)", "4. Use cram_ml() with Custom Functions" ], "created": "2025-03-27 06:23:48", "modified": "2025-04-29 14:52:57", "commits": 9 }, { "source": "cram_policy_part_2.Rmd", "filename": "cram_policy_part_2.html", "title": "Cram Policy part 2", "engine": "knitr::rmarkdown", "headings": [ "Cram Policy", "Built-in Model", "How to set model_params?", "Case of categorical target Y", "Custom Model", "1. custom_fit(X, Y, D, ...)", "2. custom_predict(model, X_new, D_new)", "3. Use cram_policy() with custom_fit() and custom_predict()", "References" ], "created": "2025-04-21 04:06:49", "modified": "2025-04-29 14:52:57", "commits": 6 }, { "source": "cram_policy_simulation.Rmd", "filename": "cram_policy_simulation.html", "title": "Cram Policy Simulation", "engine": "knitr::rmarkdown", "headings": [ "🎯 What is cram_simulation()?", "πŸ“¦ Inputs Overview", "πŸ“˜ Example: Cram Policy Simulation", "πŸ“Š Output Summary" ], "created": "2025-03-27 06:23:48", "modified": "2025-04-29 14:52:57", "commits": 8 }, { "source": "cram_policy_part_1.Rmd", "filename": "cram_policy_part_1.html", "title": "Introduction & Cram Policy part 1", "engine": "knitr::rmarkdown", "headings": [ "Introduction: What is the Cram Method?", "The Cram Workflow", "The cram_policy() Function", "πŸ”‘ Key Features of cram_policy()", "Example: Running Cram Policy on simulated data", "Interpreting Results", "Visual Summary and Notes", "πŸ“ Comparing Evaluation Strategies: Sample-Splitting, Cross-Validation, and Cram", "πŸ” Sample-Splitting", "πŸ”„ Cross-Validation", "πŸ“ˆ Cram", "🧠 Key Distinction Summary: What Is Being Evaluated?", "References" ], "created": "2025-04-21 04:06:49", "modified": "2025-05-05 19:41:33", "commits": 11 }, { "source": "quickstart.Rmd", "filename": "quickstart.html", "title": "Quick Start with CRAM", "engine": "knitr::rmarkdown", "headings": [ "Introduction", "Cram User file", "1. cram_policy() β€” Binary Policy Learning & Evaluation", "Generate Simulated Data", "Run cram_policy() with causal forest", "Case of categorical target Y", "Custom Models with cram_policy()", "2. cram_ml() β€” ML Learning & Evaluation", "Regression with cram_ml()", "Classification with cram_ml()", "Case 1: Predicting Class labels", "Case 2: Predicting Class Probabilities", "3. cram_bandit() β€” Contextual Bandits for On-policy Statistical Evaluation", "Summary" ], "created": "2025-04-02 04:48:26", "modified": "2025-04-29 14:52:57", "commits": 18 } ], "_score": 5.556302500767287, "_indexed": true, "_nocasepkg": "cramr", "_universes": [ "yanisvdc" ], "_binaries": [ { "r": "4.7.0", "os": "linux", "version": "0.1.0", "date": "2026-06-09T08:50:28.000Z", "distro": "noble", "commit": "b6d7444fd9f0ea7eabec603642b62818caf8e724", "fileid": "6edcd52c8f6e7a8bea3740d6fde661d3ee5c5cb224083cd6e9d90a01138ac397", "status": "success", "check": "OK", "buildurl": "https://github.com/r-universe/yanisvdc/actions/runs/27194410887" }, { "r": "4.6.0", "os": "linux", "version": "0.1.0", "date": "2026-06-09T08:51:12.000Z", "distro": "noble", "commit": "b6d7444fd9f0ea7eabec603642b62818caf8e724", "fileid": "e093e28a1a44648f7f4b2002a69f7f90b6743ef7bed8e18d7137185b306007e9", "status": "success", "check": "OK", "buildurl": "https://github.com/r-universe/yanisvdc/actions/runs/27194410887" }, { "r": "4.5.3", "os": "mac", "version": "0.1.0", "date": "2026-06-09T08:50:50.000Z", "commit": "b6d7444fd9f0ea7eabec603642b62818caf8e724", "fileid": "f47b07d6ceabb23c11095a6d6b7448b4730b76c082c95fa8517c07f02af490a3", "status": "success", "check": "OK", "buildurl": "https://github.com/r-universe/yanisvdc/actions/runs/27194410887" }, { "r": "4.6.0", "os": "mac", "version": "0.1.0", "date": "2026-06-09T08:49:45.000Z", "commit": "b6d7444fd9f0ea7eabec603642b62818caf8e724", "fileid": "286fe660f58666435338fcf4bea6b79e721890d121b8ef3bde34d560bcd30fd0", "status": "success", "check": "OK", "buildurl": "https://github.com/r-universe/yanisvdc/actions/runs/27194410887" }, { "r": "4.6.0", "os": "wasm", "version": "0.1.0", "date": "2026-06-09T08:50:52.000Z", "commit": "b6d7444fd9f0ea7eabec603642b62818caf8e724", "fileid": "9968ed595ef0978b229d4cdfba2908a7694c3ff3f72c9ed5b64c6b9b5cd5cc80", "status": "success", "buildurl": "https://github.com/r-universe/yanisvdc/actions/runs/27194410887" }, { "r": "4.7.0", "os": "win", "version": "0.1.0", "date": "2026-06-09T08:49:21.000Z", "commit": "b6d7444fd9f0ea7eabec603642b62818caf8e724", "fileid": "80d484957c084c8ee4231d466b048013c81c85e41e8aa8889f0c30a085386611", "status": "success", "check": "OK", "buildurl": "https://github.com/r-universe/yanisvdc/actions/runs/27194410887" }, { "r": "4.5.3", "os": "win", "version": "0.1.0", "date": "2026-06-09T08:49:16.000Z", "commit": "b6d7444fd9f0ea7eabec603642b62818caf8e724", "fileid": "93662b8d30ea0588ef49e6147806481d97753b462ea1605069f2cfa4af5689bc", "status": "success", "check": "OK", "buildurl": "https://github.com/r-universe/yanisvdc/actions/runs/27194410887" }, { "r": "4.6.0", "os": "win", "version": "0.1.0", "date": "2026-06-09T08:49:22.000Z", "commit": "b6d7444fd9f0ea7eabec603642b62818caf8e724", "fileid": "c3f87f7b8f79aa1feaa2c4962ea725cfeac14cf9b13e5c6706cc796905a7fdb7", "status": "success", "check": "OK", "buildurl": "https://github.com/r-universe/yanisvdc/actions/runs/27194410887" } ] }