post-analysis-toolkit 1.0.2__tar.gz → 1.0.4__tar.gz

{post_analysis_toolkit-1.0.2/post_analysis_toolkit.egg-info → post_analysis_toolkit-1.0.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: post-analysis-toolkit
-Version: 1.0.2
+Version: 1.0.4
 Summary: Reusable helpers for post-analysis, quasi-causal designs, and diagnostics.
 Author: OpenAI Codex
 Requires-Python: >=3.11
@@ -1151,4 +1151,4 @@ its_result = fit_its(
 
 ## Версия
 
-Текущая версия пакета: `1.0.2`
+Текущая версия пакета: `1.0.4`

{post_analysis_toolkit-1.0.2 → post_analysis_toolkit-1.0.4}/notebooks/did_its_template.ipynb

@@ -34,10 +34,10 @@
    "id": "979f6a5a",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:50.654235Z",
-     "iopub.status.busy": "2026-08-18T18:11:50.654132Z",
-     "iopub.status.idle": "2026-08-18T18:11:53.158388Z",
-     "shell.execute_reply": "2026-08-18T18:11:53.157694Z"
+     "iopub.execute_input": "2026-08-19T05:10:57.124787Z",
+     "iopub.status.busy": "2026-08-19T05:10:57.124594Z",
+     "iopub.status.idle": "2026-08-19T05:11:00.044911Z",
+     "shell.execute_reply": "2026-08-19T05:11:00.044081Z"
     }
    },
    "outputs": [],
@@ -106,10 +106,10 @@
    "id": "187f6644",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:53.160094Z",
-     "iopub.status.busy": "2026-08-18T18:11:53.159864Z",
-     "iopub.status.idle": "2026-08-18T18:11:53.173591Z",
-     "shell.execute_reply": "2026-08-18T18:11:53.173100Z"
+     "iopub.execute_input": "2026-08-19T05:11:00.046648Z",
+     "iopub.status.busy": "2026-08-19T05:11:00.046396Z",
+     "iopub.status.idle": "2026-08-19T05:11:00.061703Z",
+     "shell.execute_reply": "2026-08-19T05:11:00.061146Z"
     }
    },
    "outputs": [
@@ -210,10 +210,10 @@
    "id": "52678d87",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:53.174840Z",
-     "iopub.status.busy": "2026-08-18T18:11:53.174679Z",
-     "iopub.status.idle": "2026-08-18T18:11:53.584409Z",
-     "shell.execute_reply": "2026-08-18T18:11:53.583776Z"
+     "iopub.execute_input": "2026-08-19T05:11:00.063608Z",
+     "iopub.status.busy": "2026-08-19T05:11:00.063363Z",
+     "iopub.status.idle": "2026-08-19T05:11:03.162282Z",
+     "shell.execute_reply": "2026-08-19T05:11:03.161245Z"
     }
    },
    "outputs": [
@@ -323,6 +323,24 @@
      },
      "metadata": {},
      "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/plain": [
+       "<Figure size 1800x400 with 3 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/plain": [
+       "<Figure size 1000x400 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
     }
    ],
    "source": [
@@ -336,7 +354,9 @@
     "    unit_col=\"pizzeria\",\n",
     "    fpr_placebo_count=200,\n",
     ")\n",
-    "display(show_checks(did_diagnostics.checks))"
+    "display(show_checks(did_diagnostics.checks))\n",
+    "display(did_diagnostics.figures[\"pretrend_triptych\"])\n",
+    "display(did_diagnostics.figures[\"event_study\"])"
    ]
   },
   {
@@ -345,10 +365,10 @@
    "id": "c8f0a026",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:53.585824Z",
-     "iopub.status.busy": "2026-08-18T18:11:53.585695Z",
-     "iopub.status.idle": "2026-08-18T18:11:53.619227Z",
-     "shell.execute_reply": "2026-08-18T18:11:53.618524Z"
+     "iopub.execute_input": "2026-08-19T05:11:03.164560Z",
+     "iopub.status.busy": "2026-08-19T05:11:03.164357Z",
+     "iopub.status.idle": "2026-08-19T05:11:03.237609Z",
+     "shell.execute_reply": "2026-08-19T05:11:03.236585Z"
     }
    },
    "outputs": [
@@ -1527,6 +1547,15 @@
      },
      "metadata": {},
      "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/plain": [
+       "<Figure size 1200x400 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
     }
    ],
    "source": [
@@ -1541,7 +1570,8 @@
     "    **MDE_SETTINGS,\n",
     ")\n",
     "display(show_summary(did_result.summary))\n",
-    "display(show_frame(did_result.details[\"coefficients\"]))"
+    "display(show_frame(did_result.details[\"coefficients\"]))\n",
+    "display(did_result.figures[\"did_effect\"])"
    ]
   },
   {
@@ -1560,10 +1590,10 @@
    "id": "6ff8ba4a",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:53.621238Z",
-     "iopub.status.busy": "2026-08-18T18:11:53.621084Z",
-     "iopub.status.idle": "2026-08-18T18:11:53.624363Z",
-     "shell.execute_reply": "2026-08-18T18:11:53.623869Z"
+     "iopub.execute_input": "2026-08-19T05:11:03.239531Z",
+     "iopub.status.busy": "2026-08-19T05:11:03.239433Z",
+     "iopub.status.idle": "2026-08-19T05:11:03.242564Z",
+     "shell.execute_reply": "2026-08-19T05:11:03.241766Z"
     }
    },
    "outputs": [
@@ -1599,10 +1629,10 @@
    "id": "523adfbb",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:53.626301Z",
-     "iopub.status.busy": "2026-08-18T18:11:53.626183Z",
-     "iopub.status.idle": "2026-08-18T18:11:53.728843Z",
-     "shell.execute_reply": "2026-08-18T18:11:53.728381Z"
+     "iopub.execute_input": "2026-08-19T05:11:03.244624Z",
+     "iopub.status.busy": "2026-08-19T05:11:03.244433Z",
+     "iopub.status.idle": "2026-08-19T05:11:03.528854Z",
+     "shell.execute_reply": "2026-08-19T05:11:03.528266Z"
     }
    },
    "outputs": [
@@ -1729,10 +1759,10 @@
    "id": "87833fe4",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:53.730596Z",
-     "iopub.status.busy": "2026-08-18T18:11:53.730458Z",
-     "iopub.status.idle": "2026-08-18T18:11:53.736414Z",
-     "shell.execute_reply": "2026-08-18T18:11:53.735867Z"
+     "iopub.execute_input": "2026-08-19T05:11:03.531370Z",
+     "iopub.status.busy": "2026-08-19T05:11:03.531223Z",
+     "iopub.status.idle": "2026-08-19T05:11:03.537615Z",
+     "shell.execute_reply": "2026-08-19T05:11:03.536709Z"
     }
    },
    "outputs": [
@@ -1810,10 +1840,10 @@
    "id": "ebf028ab",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:53.738152Z",
-     "iopub.status.busy": "2026-08-18T18:11:53.737945Z",
-     "iopub.status.idle": "2026-08-18T18:11:56.847453Z",
-     "shell.execute_reply": "2026-08-18T18:11:56.846755Z"
+     "iopub.execute_input": "2026-08-19T05:11:03.539639Z",
+     "iopub.status.busy": "2026-08-19T05:11:03.539467Z",
+     "iopub.status.idle": "2026-08-19T05:11:06.924485Z",
+     "shell.execute_reply": "2026-08-19T05:11:06.923547Z"
     }
    },
    "outputs": [
@@ -1951,6 +1981,15 @@
      },
      "metadata": {},
      "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/plain": [
+       "<Figure size 1200x400 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
     }
    ],
    "source": [
@@ -1961,7 +2000,8 @@
     "    intervention_time=INTERVENTION_TIME,\n",
     "    fpr_placebo_count=200,\n",
     ")\n",
-    "display(show_checks(its_diagnostics.checks))"
+    "display(show_checks(its_diagnostics.checks))\n",
+    "display(its_diagnostics.figures[\"pre_fit\"])"
    ]
   },
   {
@@ -1970,10 +2010,10 @@
    "id": "50affca8",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:56.848757Z",
-     "iopub.status.busy": "2026-08-18T18:11:56.848655Z",
-     "iopub.status.idle": "2026-08-18T18:11:56.919201Z",
-     "shell.execute_reply": "2026-08-18T18:11:56.918600Z"
+     "iopub.execute_input": "2026-08-19T05:11:06.925931Z",
+     "iopub.status.busy": "2026-08-19T05:11:06.925795Z",
+     "iopub.status.idle": "2026-08-19T05:11:06.996372Z",
+     "shell.execute_reply": "2026-08-19T05:11:06.995573Z"
     }
    },
    "outputs": [
@@ -2257,10 +2297,10 @@
    "id": "9c501c9b",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:56.920545Z",
-     "iopub.status.busy": "2026-08-18T18:11:56.920432Z",
-     "iopub.status.idle": "2026-08-18T18:11:56.922989Z",
-     "shell.execute_reply": "2026-08-18T18:11:56.922573Z"
+     "iopub.execute_input": "2026-08-19T05:11:06.998607Z",
+     "iopub.status.busy": "2026-08-19T05:11:06.998379Z",
+     "iopub.status.idle": "2026-08-19T05:11:07.002063Z",
+     "shell.execute_reply": "2026-08-19T05:11:07.001109Z"
     }
    },
    "outputs": [
@@ -2296,10 +2336,10 @@
    "id": "5bd00664",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:56.924309Z",
-     "iopub.status.busy": "2026-08-18T18:11:56.924166Z",
-     "iopub.status.idle": "2026-08-18T18:11:57.193944Z",
-     "shell.execute_reply": "2026-08-18T18:11:57.193385Z"
+     "iopub.execute_input": "2026-08-19T05:11:07.004105Z",
+     "iopub.status.busy": "2026-08-19T05:11:07.003957Z",
+     "iopub.status.idle": "2026-08-19T05:11:07.272721Z",
+     "shell.execute_reply": "2026-08-19T05:11:07.271568Z"
     }
    },
    "outputs": [
@@ -2470,10 +2510,10 @@
    "id": "015ecfd6",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:57.195405Z",
-     "iopub.status.busy": "2026-08-18T18:11:57.195291Z",
-     "iopub.status.idle": "2026-08-18T18:11:59.704930Z",
-     "shell.execute_reply": "2026-08-18T18:11:59.704063Z"
+     "iopub.execute_input": "2026-08-19T05:11:07.274753Z",
+     "iopub.status.busy": "2026-08-19T05:11:07.274601Z",
+     "iopub.status.idle": "2026-08-19T05:11:09.908425Z",
+     "shell.execute_reply": "2026-08-19T05:11:09.907535Z"
     }
    },
    "outputs": [
@@ -2597,6 +2637,15 @@
      },
      "metadata": {},
      "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/plain": [
+       "<Figure size 1200x400 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
     }
    ],
    "source": [
@@ -2610,7 +2659,8 @@
     "    max_controls=4,\n",
     "    fpr_placebo_count=200,\n",
     ")\n",
-    "display(show_checks(causal_impact_assumptions.checks))"
+    "display(show_checks(causal_impact_assumptions.checks))\n",
+    "display(causal_impact_assumptions.figures[\"pre_fit\"])"
    ]
   },
   {
@@ -2619,10 +2669,10 @@
    "id": "f71db402",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:59.706397Z",
-     "iopub.status.busy": "2026-08-18T18:11:59.706262Z",
-     "iopub.status.idle": "2026-08-18T18:11:59.784124Z",
-     "shell.execute_reply": "2026-08-18T18:11:59.783600Z"
+     "iopub.execute_input": "2026-08-19T05:11:09.910444Z",
+     "iopub.status.busy": "2026-08-19T05:11:09.910270Z",
+     "iopub.status.idle": "2026-08-19T05:11:09.990487Z",
+     "shell.execute_reply": "2026-08-19T05:11:09.989567Z"
     }
    },
    "outputs": [
@@ -2916,10 +2966,10 @@
    "id": "2eec09cf",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:59.785431Z",
-     "iopub.status.busy": "2026-08-18T18:11:59.785313Z",
-     "iopub.status.idle": "2026-08-18T18:11:59.788363Z",
-     "shell.execute_reply": "2026-08-18T18:11:59.787576Z"
+     "iopub.execute_input": "2026-08-19T05:11:09.992523Z",
+     "iopub.status.busy": "2026-08-19T05:11:09.992414Z",
+     "iopub.status.idle": "2026-08-19T05:11:09.995718Z",
+     "shell.execute_reply": "2026-08-19T05:11:09.995173Z"
     }
    },
    "outputs": [
@@ -2956,10 +3006,10 @@
    "id": "eb6ee5b7",
    "metadata": {
     "execution": {
-     "iopub.execute_input": "2026-08-18T18:11:59.790020Z",
-     "iopub.status.busy": "2026-08-18T18:11:59.789874Z",
-     "iopub.status.idle": "2026-08-18T18:12:00.604892Z",
-     "shell.execute_reply": "2026-08-18T18:12:00.604198Z"
+     "iopub.execute_input": "2026-08-19T05:11:09.997830Z",
+     "iopub.status.busy": "2026-08-19T05:11:09.997647Z",
+     "iopub.status.idle": "2026-08-19T05:11:10.919476Z",
+     "shell.execute_reply": "2026-08-19T05:11:10.918196Z"
     }
    },
    "outputs": [

{post_analysis_toolkit-1.0.2 → post_analysis_toolkit-1.0.4}/post_analysis_toolkit/README.md

@@ -1131,4 +1131,4 @@ its_result = fit_its(
 
 ## Версия
 
-Текущая версия пакета: `1.0.2`
+Текущая версия пакета: `1.0.4`

{post_analysis_toolkit-1.0.2 → post_analysis_toolkit-1.0.4}/post_analysis_toolkit/__init__.py

@@ -2,7 +2,7 @@ from post_analysis_toolkit.compat import ensure_runtime_compat
 
 ensure_runtime_compat()
 
-__version__ = "1.0.2"
+__version__ = "1.0.4"
 
 from post_analysis_toolkit.causal_impact import (
     check_causal_impact_assumptions,

{post_analysis_toolkit-1.0.2 → post_analysis_toolkit-1.0.4}/post_analysis_toolkit/causal_impact.py

@@ -3,6 +3,7 @@ from __future__ import annotations
 from collections.abc import Iterable
 from contextlib import redirect_stdout
 import io
+import warnings
 
 import numpy as np
 import pandas as pd
@@ -159,6 +160,8 @@ def _causal_impact_placebo_fpr(
     alpha: float = 0.05,
 ) -> tuple[int, int]:
     dates = sorted(pd.to_datetime(df[time_col]).drop_duplicates())
+    if placebo_count <= 0:
+        return 0, 0
     pre_dates = [date for date in dates if date < intervention_time]
     placebo_dates = [
         date for date in pre_dates
@@ -189,6 +192,7 @@ def _causal_impact_placebo_fpr(
             upper = float(result.summary.iloc[0]["average_abs_effect_upper"])
             interval_low, interval_high = sorted((lower, upper)) if np.isfinite(lower) and np.isfinite(upper) else (np.nan, np.nan)
             significant = (np.isfinite(tail_area) and tail_area < alpha) or (np.isfinite(interval_low) and np.isfinite(interval_high) and (interval_low > 0 or interval_high < 0))
+            close_method_result_figures(result)
             if np.isfinite(tail_area) or (np.isfinite(lower) and np.isfinite(upper)):
                 runs += 1
                 false_positives += int(significant)
@@ -430,8 +434,20 @@ def fit_causal_impact(
     _ensure_nonempty_analysis_periods(ci_df, pre_period, post_period)
 
     validation_stats = _split_pre_validation(ci_df, pre_period)
-    ci_model = CausalImpact(ci_df, list(pre_period), list(post_period))
-    ci_model.run()
+    # statsmodels' HP filter emits a harmless SparseEfficiencyWarning on
+    # Databricks' scipy/statsmodels combinations.  CausalImpact does not
+    # expose the HP-filter matrix, so there is no actionable input here;
+    # suppress only this narrowly identified warning around the third-party
+    # model execution (including placebo/robustness calls through this
+    # function).
+    with warnings.catch_warnings():
+        warnings.filterwarnings(
+            "ignore",
+            message=r"spsolve requires A be CSC or CSR matrix format",
+            module=r"statsmodels\.tsa\.filters\.hp_filter",
+        )
+        ci_model = CausalImpact(ci_df, list(pre_period), list(post_period))
+        ci_model.run()
     summary = _extract_ci_summary(ci_model, post_period)
     post_obs = int(((ci_df.index >= pd.Timestamp(post_period[0])) & (ci_df.index <= pd.Timestamp(post_period[1]))).sum())
     baseline = float(ci_df.loc[(ci_df.index >= pd.Timestamp(pre_period[0])) & (ci_df.index <= pd.Timestamp(pre_period[1])), "y"].mean())

{post_analysis_toolkit-1.0.2 → post_analysis_toolkit-1.0.4}/post_analysis_toolkit/did.py

@@ -43,7 +43,7 @@ from post_analysis_toolkit.utils import (
     safe_corr,
     restrict_to_pre_period_window,
 )
-from post_analysis_toolkit.visualization import plot_estimate_series, plot_event_study, plot_pretrend_triptych
+from post_analysis_toolkit.visualization import plot_did_effect, plot_estimate_series, plot_event_study, plot_pretrend_triptych
 
 
 def _prepare_did_data(
@@ -420,6 +420,26 @@ def fit_did_regression(
         .sort_values("term")
         .reset_index(drop=True)
     )
+    grouped = (
+        work_df.assign(_time=pd.to_datetime(work_df[time_col]))
+        .groupby(["_time", treated_col], as_index=False)[analysis_outcome_col]
+        .mean()
+    )
+    treated_series = grouped.loc[grouped[treated_col].astype(int) == 1, ["_time", analysis_outcome_col]].rename(columns={analysis_outcome_col: "treated_actual"})
+    control_series = grouped.loc[grouped[treated_col].astype(int) == 0, ["_time", analysis_outcome_col]].rename(columns={analysis_outcome_col: "control_actual"})
+    did_series = treated_series.merge(control_series, on="_time", how="outer").sort_values("_time")
+    intervention_ts = pd.Timestamp(work_df.loc[work_df[post_col].astype(int) == 1, time_col].min()) if post_obs else None
+    if intervention_ts is not None:
+        treated_pre = float(did_series.loc[did_series["_time"] < intervention_ts, "treated_actual"].mean())
+        control_pre = float(did_series.loc[did_series["_time"] < intervention_ts, "control_actual"].mean())
+        did_series["treated_counterfactual"] = np.where(
+            did_series["_time"] >= intervention_ts,
+            treated_pre + did_series["control_actual"] - control_pre,
+            np.nan,
+        )
+    else:
+        did_series["treated_counterfactual"] = np.nan
+    did_series = did_series.rename(columns={"_time": time_col})
     return MethodResult(
         method=label,
         summary=summary_df,
@@ -428,6 +448,14 @@ def fit_did_regression(
             "analysis_frame": work_df,
             "mde_power_curve": mde["power_curve"],
         },
+        figures={
+            "did_effect": plot_did_effect(
+                did_series,
+                time_col=time_col,
+                intervention_time=intervention_ts,
+                title=f"{label} | {outcome_col}",
+            )
+        },
         metadata={
             "fit": fit,
             "mde_mode": mde_mode,
@@ -827,6 +855,8 @@ def _did_placebo_fpr(
 ) -> tuple[int, int]:
     pre_df = work_df.loc[work_df[time_col] < intervention_time].copy()
     dates = sorted(pre_df[time_col].drop_duplicates())
+    if not dates or placebo_count <= 0:
+        return 0, 0
     placebo_dates = [date for date in dates if date >= dates[0] + pd.Timedelta(days=28)]
     if placebo_dates:
         indexes = np.linspace(0, len(placebo_dates) - 1, min(placebo_count, len(placebo_dates)), dtype=int)
@@ -849,6 +879,7 @@ def _did_placebo_fpr(
                 mde_mode="none",
             )
             pvalues.append(float(result.summary.iloc[0]["p_value"]))
+            close_method_result_figures(result)
         except Exception:
             continue
     return len(pvalues), int(sum(p < alpha for p in pvalues))

{post_analysis_toolkit-1.0.2 → post_analysis_toolkit-1.0.4}/post_analysis_toolkit/its.py

@@ -208,6 +208,8 @@ def _its_placebo_fpr(
 ) -> tuple[int, int]:
     pre_df = work_df.loc[work_df[time_col] < intervention_time].copy()
     dates = sorted(pre_df[time_col].drop_duplicates())
+    if not dates or placebo_count <= 0:
+        return 0, 0
     placebo_dates = [
         date for date in dates
         if date >= dates[0] + pd.Timedelta(days=42)
@@ -230,6 +232,7 @@ def _its_placebo_fpr(
             )
             coefficients = result.details["coefficients"]
             level = coefficients.loc[coefficients["term"] == "post_flag", "p_value"]
+            close_method_result_figures(result)
             if level.empty:
                 continue
             runs += 1

{post_analysis_toolkit-1.0.2 → post_analysis_toolkit-1.0.4}/post_analysis_toolkit/visualization.py

@@ -95,6 +95,34 @@ def plot_actual_vs_counterfactual(
     return fig
 
 
+def plot_did_effect(
+    series_df: pd.DataFrame,
+    time_col: str,
+    treated_actual_col: str = "treated_actual",
+    control_actual_col: str = "control_actual",
+    treated_counterfactual_col: str = "treated_counterfactual",
+    intervention_time: str | pd.Timestamp | None = None,
+    title: str = "DiD effect",
+):
+    """Plot DiD group means, treated counterfactual, and post-period gap."""
+    plot_df = series_df.sort_values(time_col).copy()
+    fig, ax = plt.subplots(1, 1, figsize=(12, 4))
+    ax.plot(plot_df[time_col], plot_df[treated_actual_col], label="treated actual", linewidth=2)
+    ax.plot(plot_df[time_col], plot_df[control_actual_col], label="control actual", linewidth=2)
+    post_mask = plot_df[treated_counterfactual_col].notna()
+    ax.plot(plot_df.loc[post_mask, time_col], plot_df.loc[post_mask, treated_counterfactual_col], label="treated counterfactual", linestyle="--", linewidth=2)
+    ax.fill_between(plot_df.loc[post_mask, time_col], plot_df.loc[post_mask, treated_actual_col], plot_df.loc[post_mask, treated_counterfactual_col], alpha=0.2, label="DiD effect")
+    if intervention_time is not None:
+        ax.axvline(pd.Timestamp(intervention_time), color="black", linestyle="--", linewidth=1)
+    ax.set_title(title)
+    ax.set_xlabel(time_col)
+    ax.set_ylabel("outcome")
+    ax.grid(alpha=0.25)
+    ax.legend()
+    plt.tight_layout()
+    return fig
+
+
 def plot_gap_series(
     gap_df: pd.DataFrame,
     time_col: str,

{post_analysis_toolkit-1.0.2 → post_analysis_toolkit-1.0.4/post_analysis_toolkit.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: post-analysis-toolkit
-Version: 1.0.2
+Version: 1.0.4
 Summary: Reusable helpers for post-analysis, quasi-causal designs, and diagnostics.
 Author: OpenAI Codex
 Requires-Python: >=3.11
@@ -1151,4 +1151,4 @@ its_result = fit_its(
 
 ## Версия
 
-Текущая версия пакета: `1.0.2`
+Текущая версия пакета: `1.0.4`

{post_analysis_toolkit-1.0.2 → post_analysis_toolkit-1.0.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "post-analysis-toolkit"
-version = "1.0.2"
+version = "1.0.4"
 description = "Reusable helpers for post-analysis, quasi-causal designs, and diagnostics."
 readme = "post_analysis_toolkit/README.md"
 requires-python = ">=3.11"