lifejacket 0.2.0__py3-none-any.whl → 1.0.0__py3-none-any.whl

lifejacket/small_sample_corrections.py

@@ -16,10 +16,10 @@ logging.basicConfig(
 
 def perform_desired_small_sample_correction(
     small_sample_correction,
-    per_user_joint_adaptive_meat_contributions,
-    per_user_classical_meat_contributions,
-    per_user_classical_bread_inverse_contributions,
-    num_users,
+    per_subject_joint_adjusted_meat_contributions,
+    per_subject_classical_meat_contributions,
+    per_subject_classical_bread_inverse_contributions,
+    num_subjects,
     theta_dim,
 ):
 
@@ -28,7 +28,7 @@ def perform_desired_small_sample_correction(
     # solving linear systems (discussed more below), we typically don't have
     # issues with the conditioning of just the classical bread.
     classical_bread_inverse_matrix = jnp.mean(
-        per_user_classical_bread_inverse_contributions, axis=0
+        per_subject_classical_bread_inverse_contributions, axis=0
     )
     classical_bread_matrix = invert_matrix_and_check_conditioning(
         classical_bread_inverse_matrix,
@@ -36,60 +36,62 @@ def perform_desired_small_sample_correction(
 
     # These will hold either corrective matrices or scalar weights depending on
     # what small sample correction is requested.
-    per_user_adaptive_corrections = None
-    per_user_classical_corrections = None
+    per_subject_adjusted_corrections = None
+    per_subject_classical_corrections = None
 
-    per_user_adaptive_correction_weights = np.ones(num_users)
-    per_user_classical_correction_weights = np.ones(num_users)
+    per_subject_adjusted_correction_weights = np.ones(num_subjects)
+    per_subject_classical_correction_weights = np.ones(num_subjects)
     if small_sample_correction == SmallSampleCorrections.NONE:
         logger.info(
-            "No small sample correction requested. Using the raw per-user joint adaptive bread inverse contributions."
+            "No small sample correction requested. Using the raw per-subject joint adjusted bread inverse contributions."
         )
     elif small_sample_correction == SmallSampleCorrections.HC1theta:
         logger.info(
-            "Using HC1 small sample correction at the user trajectory level. Note that we are treating the number of parameters as simply the size of theta, despite the presence of betas."
-        )
-        per_user_adaptive_correction_weights = per_user_classical_correction_weights = (
-            num_users / (num_users - theta_dim) * np.ones(num_users)
+            "Using HC1 small sample correction at the subject trajectory level. Note that we are treating the number of parameters as simply the size of theta, despite the presence of betas."
         )
+        per_subject_adjusted_correction_weights = (
+            per_subject_classical_correction_weights
+        ) = (num_subjects / (num_subjects - theta_dim) * np.ones(num_subjects))
     elif small_sample_correction in {
         SmallSampleCorrections.HC2theta,
         SmallSampleCorrections.HC3theta,
     }:
-        logger.info("Using %s small sample correction at the user trajectory level.")
+        logger.info("Using %s small sample correction at the subject trajectory level.")
 
         power = 1 if small_sample_correction == SmallSampleCorrections.HC2theta else 2
 
-        # It turns out to typically not make sense to compute the adaptive analog
-        # of the classical leverages, since this involves correcting the joint adaptive meat matrix
+        # It turns out to typically not make sense to compute the adjusted analog
+        # of the classical leverages, since this involves correcting the joint adjusted meat matrix
         # involving all beta and theta parameters.  HC2/HC3 corrections assume that
-        # the number of parameters is smaller than the number of users, which will not typically be
-        # the case if the number of users is small enough for these corrections to be important.
-        # Therefore we also use the "classical" leverages for the adaptive correction weights, which
+        # the number of parameters is smaller than the number of subjects, which will not typically be
+        # the case if the number of subjects is small enough for these corrections to be important.
+        # Therefore we also use the "classical" leverages for the adjusted correction weights, which
         # is sensible, corresponding to only adjusting based on the estimating equations for theta.
 
         # ALSO note that one way to test correctness of the leverages is that they should sum
         # to the number of inference parameters, ie the size of theta.  I tested that this is
-        # true both for the classical leverages and the larger joint adaptive leverages when they
+        # true both for the classical leverages and the larger joint adjusted leverages when they
         # were still used, lending credence to the below calculations.
 
         # TODO: Write a unit test for some level of logic here and then rewrite this to not require
         # the classical bread explicitly. May be slower, probably needs a for loop so that can use
         # a solver for each matrix multiplication after a QR decomposition of the bread inverse
         # transpose.
-        classical_leverages_per_user = (
+        classical_leverages_per_subject = (
             np.einsum(
                 "nij,ji->n",
-                per_user_classical_bread_inverse_contributions,
+                per_subject_classical_bread_inverse_contributions,
                 classical_bread_matrix,
             )
-            / num_users
+            / num_subjects
         )
-        per_user_classical_correction_weights = 1 / (
-            (1 - classical_leverages_per_user) ** power
+        per_subject_classical_correction_weights = 1 / (
+            (1 - classical_leverages_per_subject) ** power
         )
 
-        per_user_adaptive_correction_weights = per_user_classical_correction_weights
+        per_subject_adjusted_correction_weights = (
+            per_subject_classical_correction_weights
+        )
     else:
         raise ValueError(
             f"Unknown small sample correction: {small_sample_correction}. "
@@ -98,28 +100,28 @@ def perform_desired_small_sample_correction(
 
     # If we used matrix corrections, they will be stored as these corrections.
     # Otherwise, store the scalar weights.
-    if per_user_adaptive_corrections is None:
-        per_user_adaptive_corrections = per_user_adaptive_correction_weights
-    if per_user_classical_corrections is None:
-        per_user_classical_corrections = per_user_classical_correction_weights
+    if per_subject_adjusted_corrections is None:
+        per_subject_adjusted_corrections = per_subject_adjusted_correction_weights
+    if per_subject_classical_corrections is None:
+        per_subject_classical_corrections = per_subject_classical_correction_weights
 
     # The scalar corrections will have computed weights that need to be applied here,
-    # whereas the matrix corrections will have been applied to the per-user
+    # whereas the matrix corrections will have been applied to the per-subject
     # contributions already.
-    joint_adaptive_meat_matrix = jnp.mean(
-        per_user_adaptive_correction_weights[:, None, None]
-        * per_user_joint_adaptive_meat_contributions,
+    joint_adjusted_meat_matrix = jnp.mean(
+        per_subject_adjusted_correction_weights[:, None, None]
+        * per_subject_joint_adjusted_meat_contributions,
         axis=0,
     )
     classical_meat_matrix = jnp.mean(
-        per_user_classical_correction_weights[:, None, None]
-        * per_user_classical_meat_contributions,
+        per_subject_classical_correction_weights[:, None, None]
+        * per_subject_classical_meat_contributions,
         axis=0,
     )
 
     return (
-        joint_adaptive_meat_matrix,
+        joint_adjusted_meat_matrix,
         classical_meat_matrix,
-        per_user_adaptive_corrections,
-        per_user_classical_corrections,
+        per_subject_adjusted_corrections,
+        per_subject_classical_corrections,
     )

lifejacket-1.0.0.dist-info/METADATA

@@ -0,0 +1,56 @@
+Metadata-Version: 2.4
+Name: lifejacket
+Version: 1.0.0
+Summary: Consistent standard errors for longitudinal data collected under pooling online decision policies.
+Author-email: Nowell Closser <nowellclosser@gmail.com>
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+Requires-Dist: click>=8.0
+Requires-Dist: jax>=0.4.0
+Requires-Dist: jaxlib>=0.4.0
+Requires-Dist: numpy>=1.20.0
+Requires-Dist: pandas>=1.3.0
+Requires-Dist: scipy>=1.7.0
+Requires-Dist: plotext>=5.0.0
+Provides-Extra: dev
+Requires-Dist: pytest>=7.0; extra == "dev"
+Requires-Dist: black>=22.0; extra == "dev"
+Requires-Dist: flake8>=4.0; extra == "dev"
+
+```python
+  _ _  __     _            _        _
+ | (_)/ _|   (_)          | |      | |
+ | |_| |_ ___ _  __ _  ___| | _____| |_
+ | | |  _/ _ \ |/ _` |/ __| |/ / _ \ __|
+ | | | ||  __/ | (_| | (__|   <  __/ |_
+ |_|_|_| \___| |\__,_|\___|_|\_\___|\__|
+            _/ |
+           |__/
+```
+
+Save your standard errors from pooling in online decision-making algorithms.
+
+## Setup (if not using conda)
+### Create and activate a virtual environment
+- `python3 -m venv .venv; source /.venv/bin/activate`
+
+### Adding a package
+- Add to `requirements.txt` with a specific version or no version if you want the latest stable
+- Run `pip freeze > requirements.txt` to lock the versions of your package and all its subpackages
+
+## Running the code
+- `export PYTHONPATH to the absolute path of this repository on your computer
+- `./run_local_synthetic.sh`, which outputs to `simulated_data/` by default. See all the possible flags to be toggled in the script code.
+
+## Linting/Formatting
+
+## Testing
+python -m pytest
+python -m pytest tests/unit_tests
+python -m pytest tests/integration_tests
+
+
+
+## TODO
+1. Add precommit hooks (pip freeze, linting, formatting)
+

lifejacket-1.0.0.dist-info/RECORD

@@ -0,0 +1,17 @@
+lifejacket/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+lifejacket/after_study_analysis.py,sha256=-6LCfl0dz9eHkP5aLs-J2YqpxVJ57J8Ov4wEWKbZMWA,83315
+lifejacket/arg_threading_helpers.py,sha256=jiPdG-1USYUXqXAEfRPL-3KBKno7uMr4-fCl9GrZ7Fw,17599
+lifejacket/calculate_derivatives.py,sha256=wFMS1pSm5MI8D7FYK2nPEAdvtVJAxKy8OVtUj7giII0,37520
+lifejacket/constants.py,sha256=2L05p6NJ7l3qRZ1hD2KlrvzWF1ReSmWRUkULPIhdvJo,842
+lifejacket/deployment_conditioning_monitor.py,sha256=5PMBfDQfyxwoYXr3qh1pTdMBefnSjjQMFyCj3_sItVY,43526
+lifejacket/form_adjusted_meat_adjustments_directly.py,sha256=ML7FOB_hnDEM1ndC2X4j_oaqdqn-EPr9j-MetlOqSE8,13740
+lifejacket/get_datum_for_blowup_supervised_learning.py,sha256=SjA3H7H8RBZc9GxzgsccqZoLZnAD68mxaPH1E87s2lA,59092
+lifejacket/helper_functions.py,sha256=ldkanFn5b2lwBlni-HtXPVd7QvxEKmzL1UsJMsG6f6g,23478
+lifejacket/input_checks.py,sha256=RhS2tnq74jOU5_pVSXsxthSmuGhlmoMc5eJiourQiS0,47513
+lifejacket/small_sample_corrections.py,sha256=bnbtxjEKZBiFrCSF2WA0vczaVeIqao1NekxJRZdmwCU,5692
+lifejacket/vmap_helpers.py,sha256=pZqYN3p9Ty9DPOeeY9TKbRJXR2AV__HBwwDFOvdOQ84,2688
+lifejacket-1.0.0.dist-info/METADATA,sha256=UGZsalPpzX8K7BnstZ0RDFq-1ZpL5CsXcUd_hYvQWus,1773
+lifejacket-1.0.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+lifejacket-1.0.0.dist-info/entry_points.txt,sha256=4k8ibVIUT-OHxPaaDv-QwWpC64ErzhdemHpTAXCnb8w,67
+lifejacket-1.0.0.dist-info/top_level.txt,sha256=vKl8m7jOQ4pkbzVuHCJsq-8LcXRrOAWnok3bBo9qpsE,11
+lifejacket-1.0.0.dist-info/RECORD,,

lifejacket-0.2.0.dist-info/METADATA

@@ -1,100 +0,0 @@
-Metadata-Version: 2.4
-Name: lifejacket
-Version: 0.2.0
-Summary: A package for after-study analysis of adaptive experiments in which data is pooled across users.
-Author-email: Nowell Closser <nowellclosser@gmail.com>
-Requires-Python: >=3.10
-Description-Content-Type: text/markdown
-Requires-Dist: click>=8.0
-Requires-Dist: jax>=0.4.0
-Requires-Dist: jaxlib>=0.4.0
-Requires-Dist: numpy>=1.20.0
-Requires-Dist: pandas>=1.3.0
-Requires-Dist: scipy>=1.7.0
-Requires-Dist: plotext>=5.0.0
-Provides-Extra: dev
-Requires-Dist: pytest>=7.0; extra == "dev"
-Requires-Dist: black>=22.0; extra == "dev"
-Requires-Dist: flake8>=4.0; extra == "dev"
-
-```python
-  _ _  __     _            _        _
- | (_)/ _|   (_)          | |      | |
- | |_| |_ ___ _  __ _  ___| | _____| |_
- | | |  _/ _ \ |/ _` |/ __| |/ / _ \ __|
- | | | ||  __/ | (_| | (__|   <  __/ |_
- |_|_|_| \___| |\__,_|\___|_|\_\___|\__|
-            _/ |
-           |__/
-```
-
-Save your standard errors from pooling in adaptive experiments.
-
-## Setup (if not using conda)
-### Create and activate a virtual environment
-- `python3 -m venv .venv; source /.venv/bin/activate`
-
-### Adding a package
-- Add to `requirements.txt` with a specific version or no version if you want the latest stable
-- Run `pip freeze > requirements.txt` to lock the versions of your package and all its subpackages
-
-## Running the code
-- `export PYTHONPATH to the absolute path of this repository on your computer
-- `./run_local_synthetic.sh`, which outputs to `simulated_data/` by default. See all the possible flags to be toggled in the script code.
-
-## Linting/Formatting
-
-## Testing
-python -m pytest
-python -m pytest tests/unit_tests
-python -m pytest tests/integration_tests
-
-
-# Talk about gitignored cluster simulation scripts
-
-
-
-
-
-
-### Important Large-Scale Simulations
-
-#### No adaptivity
-sbatch --array=[0-999] -t 0-5:00 --mem=50G run_and_analysis_parallel_synthetic --T=10 --n=50000 --recruit_n=50000 --steepness=0.0 --alg_state_feats=intercept,past_reward --action_centering_RL=0 --inference_loss_func_filename="functions_to_pass_to_analysis/get_least_squares_loss_inference_no_action_centering.py" --theta_calculation_func_filename="functions_to_pass_to_analysis/estimate_theta_least_squares_no_action_centering.py"
-
-#### No adaptivity, 5 batches incremental recruitment
-sbatch --array=[0-999] -t 0-5:00 --mem=50G run_and_analysis_parallel_synthetic --T=10 --n=50000 --recruit_n=10000 --steepness=0.0 --alg_state_feats=intercept,past_reward --action_centering_RL=0 --inference_loss_func_filename="functions_to_pass_to_analysis/get_least_squares_loss_inference_no_action_centering.py" --theta_calculation_func_filename="functions_to_pass_to_analysis/estimate_theta_least_squares_no_action_centering.py"
-
-#### Some adaptivity, no action_centering
-sbatch --array=[0-999] -t 0-5:00 --mem=50G run_and_analysis_parallel_synthetic --T=10 --n=50000 --recruit_n=50000 --steepness=3.0 --alg_state_feats=intercept,past_reward --action_centering_RL=0 --inference_loss_func_filename="functions_to_pass_to_analysis/get_least_squares_loss_inference_no_action_centering.py" --theta_calculation_func_filename="functions_to_pass_to_analysis/estimate_theta_least_squares_no_action_centering.py"
-
-#### Some adaptivity, no action_centering, 5 batches incremental recruitment
-sbatch --array=[0-999] -t 0-5:00 --mem=50G run_and_analysis_parallel_synthetic --T=10 --n=50000 --recruit_n=10000 --steepness=3.0 --alg_state_feats=intercept,past_reward --action_centering_RL=0 --inference_loss_func_filename="functions_to_pass_to_analysis/get_least_squares_loss_inference_no_action_centering.py" --theta_calculation_func_filename="functions_to_pass_to_analysis/estimate_theta_least_squares_no_action_centering.py"
-
-#### More adaptivity, no action_centering
-sbatch --array=[0-999] -t 0-5:00 --mem=50G run_and_analysis_parallel_synthetic --T=10 --n=50000 --recruit_n=50000 --steepness=5.0 --alg_state_feats=intercept,past_reward --action_centering_RL=0 --inference_loss_func_filename="functions_to_pass_to_analysis/get_least_squares_loss_inference_no_action_centering.py" --theta_calculation_func_filename="functions_to_pass_to_analysis/estimate_theta_least_squares_no_action_centering.py"
-
-#### Even more adaptivity, no action_centering
-sbatch --array=[0-999] -t 0-5:00 --mem=50G run_and_analysis_parallel_synthetic --T=10 --n=50000 --recruit_n=50000 --steepness=10.0 --alg_state_feats=intercept,past_reward --action_centering_RL=0 --inference_loss_func_filename="functions_to_pass_to_analysis/get_least_squares_loss_inference_no_action_centering.py" --theta_calculation_func_filename="functions_to_pass_to_analysis/estimate_theta_least_squares_no_action_centering.py"
-
-#### Some adaptivity, RL action_centering, no inference action centering
-sbatch --array=[0-999] -t 0-5:00 --mem=50G run_and_analysis_parallel_synthetic --T=10 --n=50000 --recruit_n=50000 --steepness=3.0 --alg_state_feats=intercept,past_reward --action_centering_RL=1 --inference_loss_func_filename="functions_to_pass_to_analysis/get_least_squares_loss_inference_no_action_centering.py" --theta_calculation_func_filename="functions_to_pass_to_analysis/estimate_theta_least_squares_no_action_centering.py"
-
-#### Some adaptivity, inference action_centering, no RL action centering
-sbatch --array=[0-999] -t 0-5:00 --mem=50G run_and_analysis_parallel_synthetic --T=10 --n=50000 --recruit_n=50000 --steepness=3.0 --alg_state_feats=intercept,past_reward --action_centering_RL=0 --inference_loss_func_filename="functions_to_pass_to_analysis/get_least_squares_loss_inference_action_centering.py" --theta_calculation_func_filename="functions_to_pass_to_analysis/estimate_theta_least_squares_action_centering.py"
-
-#### Some adaptivity, inference and RL action_centering
-sbatch --array=[0-999] -t 0-5:00 --mem=50G run_and_analysis_parallel_synthetic --T=10 --n=50000 --recruit_n=50000 --steepness=3.0 --alg_state_feats=intercept,past_reward --action_centering_RL=1 --inference_loss_func_filename="functions_to_pass_to_analysis/get_least_squares_loss_inference_action_centering.py" --theta_calculation_func_filename="functions_to_pass_to_analysis/estimate_theta_least_squares_action_centering.py"
-
-#### Some adaptivity, inference and RL action_centering, even more T
-sbatch --array=[0-999] -t 1-00:00 --mem=50G run_and_analysis_parallel_synthetic --T=25 --n=50000 --recruit_n=50000 --steepness=3.0 --alg_state_feats=intercept,past_reward --action_centering_RL=1 --inference_loss_func_filename="functions_to_pass_to_analysis/get_least_squares_loss_inference_no_action_centering.py" --theta_calculation_func_filename="functions_to_pass_to_analysis/estimate_theta_least_squares_no_action_centering.py"
-
-#### Some adaptivity, inference and RL action_centering, even more T, 5 batches incremental recruitment
-sbatch --array=[0-999] -t 1-00:00 --mem=50G run_and_analysis_parallel_synthetic --T=25 --n=50000 --recruit_n=10000 --steepness=3.0 --alg_state_feats=intercept,past_reward --action_centering_RL=1 --inference_loss_func_filename="functions_to_pass_to_analysis/get_least_squares_loss_inference_no_action_centering.py" --theta_calculation_func_filename="functions_to_pass_to_analysis/estimate_theta_least_squares_no_action_centering.py"
-
-
-
-## TODO
-1. Add precommit hooks (pip freeze, linting, formatting)
-2. Run tests on PRs
-

lifejacket-0.2.0.dist-info/RECORD

@@ -1,17 +0,0 @@
-lifejacket/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-lifejacket/after_study_analysis.py,sha256=_Weeca51EXWlQkx3IWd2t1jImBxODOTe9-8gtKBSlus,82168
-lifejacket/arg_threading_helpers.py,sha256=7HV7qkiJtm-E-cpAhtv4n_BpCVjz9tC0nGENbc090h8,17282
-lifejacket/calculate_derivatives.py,sha256=SceXFWtK56uCCdXGD7v8JijgYz0UCBKzcnrPH_nAqNE,37536
-lifejacket/constants.py,sha256=2L05p6NJ7l3qRZ1hD2KlrvzWF1ReSmWRUkULPIhdvJo,842
-lifejacket/form_adaptive_meat_adjustments_directly.py,sha256=bSLrVYLZR1-Qlm5yIdktzv8ZQTVhHTlhVVL2wEjLTmw,13737
-lifejacket/get_datum_for_blowup_supervised_learning.py,sha256=V8H4PE49dQwsKjj93QEu2BKbhwPr56QMtx2jhan39-c,58357
-lifejacket/helper_functions.py,sha256=xOhRG-Cm4ZdRNm-O0faHna583d74pLWY5_jfnokegWc,23295
-lifejacket/input_checks.py,sha256=KcDdfsdCVCKKcx07FfOKJb3KVX6xFuWwAufGJ3msAuc,46972
-lifejacket/small_sample_corrections.py,sha256=f8jmg9U9ZN77WadJud70tt6NMxCTsSGtlsdF_-mfws4,5543
-lifejacket/trial_conditioning_monitor.py,sha256=qNTHh0zY2P7zJxox_OwhEEK8Ed1l0TPOjGDqNxMNoIQ,42164
-lifejacket/vmap_helpers.py,sha256=pZqYN3p9Ty9DPOeeY9TKbRJXR2AV__HBwwDFOvdOQ84,2688
-lifejacket-0.2.0.dist-info/METADATA,sha256=XzZQXRKzohTg0FOv0sIh7Ii6Snj2m5HVIdeXVYUphe4,7274
-lifejacket-0.2.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-lifejacket-0.2.0.dist-info/entry_points.txt,sha256=4k8ibVIUT-OHxPaaDv-QwWpC64ErzhdemHpTAXCnb8w,67
-lifejacket-0.2.0.dist-info/top_level.txt,sha256=vKl8m7jOQ4pkbzVuHCJsq-8LcXRrOAWnok3bBo9qpsE,11
-lifejacket-0.2.0.dist-info/RECORD,,