predikit 0.4.3__tar.gz → 0.4.5__tar.gz

{predikit-0.4.3 → predikit-0.4.5}/CHANGELOG.md

@@ -4,7 +4,29 @@ All notable changes to this project will be documented in this file.
 
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
-## [Unreleased]
+## [Unreleased]
+
+## [0.4.5] - 2026-07-01
+
+### Fixed
+- Added validation for `confidence_threshold` so values outside `0.0` to `1.0` fail at construction time
+- Added `ModelEnsemble` validation for duplicate `collect` outputs, mismatched aggregate output names, negative weights, and zero-total weighted strategies
+- Added `ToolRegistry` duplicate-name validation across tools and ensembles
+- Updated `ToolRegistry.get()` to retrieve both tools and ensembles with clearer missing-name errors
+- Cleaned up example lint issues and corrected stale `modelbridge[xgboost]` install text
+
+### Changed
+- Marked MLflow loader tests as opt-in integration tests so the default pytest suite stays fast and deterministic
+
+## [0.4.4] - 2026-06-24
+
+### Changed
+- README overhauled: added "Why predikit?" comparison table, "Works with" ecosystem section, GitHub star/fork badges, and a shipped-vs-planned roadmap
+- Quick start section replaces "30-second example" heading; `---` dividers added between major sections for better scanability
+- PyPI package description updated to a more descriptive one-liner
+- Keywords expanded: added `scikit-learn`, `tool-use`, `openai`, `langchain`, `mlops`, `ai-agents`, `pydantic`, `model-serving`
+- Added `Documentation` URL to `[project.urls]` in `pyproject.toml`
+- Added `Topic :: Software Development :: Libraries :: Python Modules` and `Typing :: Typed` classifiers
 
 ## [0.4.2] - 2026-06-13
 

{predikit-0.4.3 → predikit-0.4.5}/PKG-INFO

@@ -1,15 +1,16 @@
 Metadata-Version: 2.4
 Name: predikit
-Version: 0.4.3
-Summary: Turn any trained sklearn/XGBoost model into an LLM-callable tool with auto-generated schemas and typed I/O.
+Version: 0.4.5
+Summary: Wrap any trained sklearn/XGBoost model as an LLM-callable tool — auto-generated JSON schemas, Pydantic I/O, OpenAI & LangChain ready, zero boilerplate.
 Project-URL: Homepage, https://github.com/Tejas-TA/predikit
 Project-URL: Repository, https://github.com/Tejas-TA/predikit
+Project-URL: Documentation, https://github.com/Tejas-TA/predikit#readme
 Project-URL: Bug Tracker, https://github.com/Tejas-TA/predikit/issues
 Project-URL: Changelog, https://github.com/Tejas-TA/predikit/releases
 Author-email: Tejas Tumakuru Ashok <tejasta@gmail.com>
 License: MIT
 License-File: LICENSE
-Keywords: agents,function-calling,llm,ml-tools,sklearn,xgboost
+Keywords: agents,ai-agents,function-calling,langchain,llm,machine-learning,ml-tools,mlops,model-serving,openai,pydantic,scikit-learn,sklearn,tool-use,xgboost
 Classifier: Development Status :: 4 - Beta
 Classifier: Intended Audience :: Developers
 Classifier: Intended Audience :: Science/Research
@@ -19,6 +20,8 @@ Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: Software Development :: Libraries :: Python Modules
+Classifier: Typing :: Typed
 Requires-Python: >=3.10
 Requires-Dist: numpy>=1.24
 Requires-Dist: pydantic>=2.0
@@ -56,7 +59,8 @@ Description-Content-Type: text/markdown
 </p>
 
 <p align="center">
-  Turn any trained scikit-learn or XGBoost model into an LLM-callable tool —<br/>
+  <strong>The bridge between your ML models and LLM agents.</strong><br/>
+  Wrap any trained scikit-learn or XGBoost model as an LLM-callable tool —<br/>
   auto-generated JSON schemas, typed I/O, zero boilerplate.
 </p>
 
@@ -68,6 +72,8 @@ Description-Content-Type: text/markdown
 </p>
 
 <p align="center">
+  <a href="https://github.com/Tejas-TA/predikit/stargazers"><img src="https://img.shields.io/github/stars/Tejas-TA/predikit?style=social" alt="GitHub Stars"/></a>
+  <a href="https://github.com/Tejas-TA/predikit/network/members"><img src="https://img.shields.io/github/forks/Tejas-TA/predikit?style=social" alt="GitHub Forks"/></a>
   <a href="https://pepy.tech/project/predikit"><img src="https://static.pepy.tech/personalized-badge/predikit?period=week&units=international_system&left_color=grey&right_color=blue&left_text=weekly+downloads" alt="Weekly Downloads"/></a>
   <a href="https://pepy.tech/project/predikit"><img src="https://static.pepy.tech/personalized-badge/predikit?period=month&units=international_system&left_color=grey&right_color=blue&left_text=monthly+downloads" alt="Monthly Downloads"/></a>
   <a href="https://pepy.tech/project/predikit"><img src="https://static.pepy.tech/personalized-badge/predikit?period=total&units=international_system&left_color=grey&right_color=blue&left_text=total+downloads" alt="Total Downloads"/></a>
@@ -79,34 +85,71 @@ tool.to_openai()              # OpenAI function schema, ready to pass to the API
 tool.invoke({"sqft": 2200})   # → {"price_usd": 370730}
 ```
 
+---
+
 ## Table of Contents
+- [Why predikit?](#why-predikit)
+- [Works with](#works-with)
 - [Install](#install)
-- [30-second example](#30-second-example)
+- [Quick start](#quick-start)
 - [Core API](#core-api)
 - [Field naming rule](#field-naming-rule)
 - [Cookbook](#cookbook)
+- [Roadmap](#roadmap)
 - [Contributing](#contributing)
 - [License](#license)
 
-## Install
+---
 
-```bash
-pip install predikit
+## Why predikit?
+
+Most ML pipelines stop at `.predict()`. Getting that prediction callable by an LLM agent — with validated inputs, correct types, and a schema the model can reason about — requires glue code that's tedious to write and easy to get wrong.
+
+predikit handles that layer for you:
+
+|  | Without predikit | With predikit |
+|--|------------------|---------------|
+| **Schema** | Hand-write JSON Schema for every model | Auto-generated from your Pydantic `BaseModel` |
+| **Type safety** | Manual casting, silent failures | Pydantic v2 validation with clear error messages |
+| **LLM integration** | OpenAI / LangChain boilerplate per model | `.to_openai()` / `.to_langchain()` in one line |
+| **Ensemble routing** | Custom aggregation logic per project | `ModelEnsemble` with 5 built-in strategies |
+| **Confidence handling** | Write your own threshold checks | `confidence_threshold` + `on_low_confidence` |
+| **Model registries** | Manual MLflow / Snowflake registry calls | `from_mlflow()` / `from_snowflake()` loaders |
+| **Async** | `asyncio.get_event_loop().run_in_executor(...)` | `await tool.ainvoke(inputs)` |
+
+---
+
+## Works with
+
+predikit is designed to plug into the tools your team already uses:
+
+**Models** — scikit-learn · XGBoost
+
+**LLM frameworks** — OpenAI function calling · LangChain · any tool-calling API that accepts JSON Schema
+
+**Model registries** — MLflow Model Registry · Snowflake Model Registry
+
+**Validation** — Pydantic v2
 
-# With XGBoost support
-pip install predikit[xgboost]
+**Async** — `asyncio`-compatible via `ainvoke()`
 
-# With LangChain support
-pip install predikit[langchain]
+---
 
-# With MLflow Model Registry support
-pip install predikit[mlflow]
+## Install
+
+```bash
+pip install predikit
 
-# With Snowflake Model Registry support
-pip install predikit[snowflake]
+# Optional extras
+pip install predikit[xgboost]    # XGBoost support
+pip install predikit[langchain]  # LangChain StructuredTool export
+pip install predikit[mlflow]     # MLflow Model Registry loader
+pip install predikit[snowflake]  # Snowflake Model Registry loader
 ```
 
-## 30-second example
+---
+
+## Quick start
 
 ```python
 from pydantic import BaseModel, Field
@@ -147,6 +190,8 @@ tool.invoke({
 # → {"species": 0}
 ```
 
+---
+
 ## Core API
 
 ### `ModelTool`
@@ -205,6 +250,8 @@ ModelEnsemble(
 
 `ModelEnsemble` exposes the same `.invoke()`, `.ainvoke()`, `.to_openai()`, and `.to_langchain()` interface as `ModelTool`.
 
+---
+
 ## Field naming rule
 
 **Your Pydantic schema field names must exactly match the column names the model was trained on.**
@@ -234,6 +281,8 @@ Schema has: ['square_footage', 'beds', 'bathrooms'], model expects: ['sqft', 'be
 
 > **Tip:** If you trained with a numpy array (no DataFrame), predikit has no feature names to check — it uses your schema's field definition order instead.
 
+---
+
 ## Cookbook
 
 ### XGBoost regression
@@ -389,23 +438,41 @@ tool = from_snowflake(
 
 Pass `output_method="predict_proba"` or any other method your Snowflake model exposes. The returned `ModelTool` is identical to one built directly — all exporters, confidence routing, and ensemble strategies work as-is. Requires `pip install predikit[snowflake]`.
 
-### Orlando real estate demo
+### End-to-end demo
+
+See [`examples/03_orlando_real_estate.py`](examples/03_orlando_real_estate.py) for a full walkthrough: synthetic dataset → XGBoost training → `ModelTool` → registry → OpenAI schema → prediction.
 
-See [`examples/03_orlando_real_estate.py`](examples/03_orlando_real_estate.py) for a full end-to-end walkthrough: synthetic dataset → XGBoost training → `ModelTool` → registry → OpenAI schema → prediction.
+---
 
 ## Roadmap
 
-Planned for later releases:
+### Shipped
+
+- [x] OpenAI function-calling schema export
+- [x] LangChain `StructuredTool` export
+- [x] Pydantic v2 typed I/O validation
+- [x] Multi-model `ModelEnsemble` — collect / mean / vote / weighted variants
+- [x] Confidence routing — `warn` / `raise` / `fallback`
+- [x] Async `ainvoke()` via thread pool executor
+- [x] MLflow Model Registry loader
+- [x] Snowflake Model Registry loader
+- [x] `predikit inspect` CLI
+
+### Planned
 
-- HuggingFace / PyTorch / TensorFlow model support
-- Streaming inference support
-- OpenAI Assistants API integration
+- [ ] HuggingFace / PyTorch / TensorFlow model support
+- [ ] Streaming inference support
+- [ ] OpenAI Assistants API integration
+- [ ] MCP server mode
+
+---
 
 ## Contributing
 
 See [CONTRIBUTING.md](CONTRIBUTING.md) for development setup, code style, and PR guidelines. The [CHANGELOG](CHANGELOG.md) tracks notable changes per release.
 
+Issues and PRs are welcome — if you're wrapping a model type or registry that predikit doesn't support yet, open a discussion.
+
 ## License
 
 MIT © Tejas Tumakuru Ashok
-

{predikit-0.4.3 → predikit-0.4.5}/README.md

@@ -6,7 +6,8 @@
 </p>
 
 <p align="center">
-  Turn any trained scikit-learn or XGBoost model into an LLM-callable tool —<br/>
+  <strong>The bridge between your ML models and LLM agents.</strong><br/>
+  Wrap any trained scikit-learn or XGBoost model as an LLM-callable tool —<br/>
   auto-generated JSON schemas, typed I/O, zero boilerplate.
 </p>
 
@@ -18,6 +19,8 @@
 </p>
 
 <p align="center">
+  <a href="https://github.com/Tejas-TA/predikit/stargazers"><img src="https://img.shields.io/github/stars/Tejas-TA/predikit?style=social" alt="GitHub Stars"/></a>
+  <a href="https://github.com/Tejas-TA/predikit/network/members"><img src="https://img.shields.io/github/forks/Tejas-TA/predikit?style=social" alt="GitHub Forks"/></a>
   <a href="https://pepy.tech/project/predikit"><img src="https://static.pepy.tech/personalized-badge/predikit?period=week&units=international_system&left_color=grey&right_color=blue&left_text=weekly+downloads" alt="Weekly Downloads"/></a>
   <a href="https://pepy.tech/project/predikit"><img src="https://static.pepy.tech/personalized-badge/predikit?period=month&units=international_system&left_color=grey&right_color=blue&left_text=monthly+downloads" alt="Monthly Downloads"/></a>
   <a href="https://pepy.tech/project/predikit"><img src="https://static.pepy.tech/personalized-badge/predikit?period=total&units=international_system&left_color=grey&right_color=blue&left_text=total+downloads" alt="Total Downloads"/></a>
@@ -29,34 +32,71 @@ tool.to_openai()              # OpenAI function schema, ready to pass to the API
 tool.invoke({"sqft": 2200})   # → {"price_usd": 370730}
 ```
 
+---
+
 ## Table of Contents
+- [Why predikit?](#why-predikit)
+- [Works with](#works-with)
 - [Install](#install)
-- [30-second example](#30-second-example)
+- [Quick start](#quick-start)
 - [Core API](#core-api)
 - [Field naming rule](#field-naming-rule)
 - [Cookbook](#cookbook)
+- [Roadmap](#roadmap)
 - [Contributing](#contributing)
 - [License](#license)
 
-## Install
+---
 
-```bash
-pip install predikit
+## Why predikit?
+
+Most ML pipelines stop at `.predict()`. Getting that prediction callable by an LLM agent — with validated inputs, correct types, and a schema the model can reason about — requires glue code that's tedious to write and easy to get wrong.
+
+predikit handles that layer for you:
+
+|  | Without predikit | With predikit |
+|--|------------------|---------------|
+| **Schema** | Hand-write JSON Schema for every model | Auto-generated from your Pydantic `BaseModel` |
+| **Type safety** | Manual casting, silent failures | Pydantic v2 validation with clear error messages |
+| **LLM integration** | OpenAI / LangChain boilerplate per model | `.to_openai()` / `.to_langchain()` in one line |
+| **Ensemble routing** | Custom aggregation logic per project | `ModelEnsemble` with 5 built-in strategies |
+| **Confidence handling** | Write your own threshold checks | `confidence_threshold` + `on_low_confidence` |
+| **Model registries** | Manual MLflow / Snowflake registry calls | `from_mlflow()` / `from_snowflake()` loaders |
+| **Async** | `asyncio.get_event_loop().run_in_executor(...)` | `await tool.ainvoke(inputs)` |
+
+---
+
+## Works with
+
+predikit is designed to plug into the tools your team already uses:
+
+**Models** — scikit-learn · XGBoost
+
+**LLM frameworks** — OpenAI function calling · LangChain · any tool-calling API that accepts JSON Schema
+
+**Model registries** — MLflow Model Registry · Snowflake Model Registry
+
+**Validation** — Pydantic v2
 
-# With XGBoost support
-pip install predikit[xgboost]
+**Async** — `asyncio`-compatible via `ainvoke()`
 
-# With LangChain support
-pip install predikit[langchain]
+---
 
-# With MLflow Model Registry support
-pip install predikit[mlflow]
+## Install
+
+```bash
+pip install predikit
 
-# With Snowflake Model Registry support
-pip install predikit[snowflake]
+# Optional extras
+pip install predikit[xgboost]    # XGBoost support
+pip install predikit[langchain]  # LangChain StructuredTool export
+pip install predikit[mlflow]     # MLflow Model Registry loader
+pip install predikit[snowflake]  # Snowflake Model Registry loader
 ```
 
-## 30-second example
+---
+
+## Quick start
 
 ```python
 from pydantic import BaseModel, Field
@@ -97,6 +137,8 @@ tool.invoke({
 # → {"species": 0}
 ```
 
+---
+
 ## Core API
 
 ### `ModelTool`
@@ -155,6 +197,8 @@ ModelEnsemble(
 
 `ModelEnsemble` exposes the same `.invoke()`, `.ainvoke()`, `.to_openai()`, and `.to_langchain()` interface as `ModelTool`.
 
+---
+
 ## Field naming rule
 
 **Your Pydantic schema field names must exactly match the column names the model was trained on.**
@@ -184,6 +228,8 @@ Schema has: ['square_footage', 'beds', 'bathrooms'], model expects: ['sqft', 'be
 
 > **Tip:** If you trained with a numpy array (no DataFrame), predikit has no feature names to check — it uses your schema's field definition order instead.
 
+---
+
 ## Cookbook
 
 ### XGBoost regression
@@ -339,23 +385,41 @@ tool = from_snowflake(
 
 Pass `output_method="predict_proba"` or any other method your Snowflake model exposes. The returned `ModelTool` is identical to one built directly — all exporters, confidence routing, and ensemble strategies work as-is. Requires `pip install predikit[snowflake]`.
 
-### Orlando real estate demo
+### End-to-end demo
+
+See [`examples/03_orlando_real_estate.py`](examples/03_orlando_real_estate.py) for a full walkthrough: synthetic dataset → XGBoost training → `ModelTool` → registry → OpenAI schema → prediction.
 
-See [`examples/03_orlando_real_estate.py`](examples/03_orlando_real_estate.py) for a full end-to-end walkthrough: synthetic dataset → XGBoost training → `ModelTool` → registry → OpenAI schema → prediction.
+---
 
 ## Roadmap
 
-Planned for later releases:
+### Shipped
+
+- [x] OpenAI function-calling schema export
+- [x] LangChain `StructuredTool` export
+- [x] Pydantic v2 typed I/O validation
+- [x] Multi-model `ModelEnsemble` — collect / mean / vote / weighted variants
+- [x] Confidence routing — `warn` / `raise` / `fallback`
+- [x] Async `ainvoke()` via thread pool executor
+- [x] MLflow Model Registry loader
+- [x] Snowflake Model Registry loader
+- [x] `predikit inspect` CLI
+
+### Planned
 
-- HuggingFace / PyTorch / TensorFlow model support
-- Streaming inference support
-- OpenAI Assistants API integration
+- [ ] HuggingFace / PyTorch / TensorFlow model support
+- [ ] Streaming inference support
+- [ ] OpenAI Assistants API integration
+- [ ] MCP server mode
+
+---
 
 ## Contributing
 
 See [CONTRIBUTING.md](CONTRIBUTING.md) for development setup, code style, and PR guidelines. The [CHANGELOG](CHANGELOG.md) tracks notable changes per release.
 
+Issues and PRs are welcome — if you're wrapping a model type or registry that predikit doesn't support yet, open a discussion.
+
 ## License
 
 MIT © Tejas Tumakuru Ashok
-

{predikit-0.4.3 → predikit-0.4.5}/examples/01_basic_sklearn.py

@@ -1,4 +1,5 @@
 """Example 01 — Basic sklearn classifier wrapped as an OpenAI tool."""
+
 import json
 
 from pydantic import BaseModel, Field
@@ -36,7 +37,9 @@ print(json.dumps(tool.to_openai(), indent=2))
 
 # 5. Direct invocation
 print("\n=== Direct Invocation ===")
-result = tool.invoke({"sepal_length": 5.1, "sepal_width": 3.5, "petal_length": 1.4, "petal_width": 0.2})
+result = tool.invoke(
+    {"sepal_length": 5.1, "sepal_width": 3.5, "petal_length": 1.4, "petal_width": 0.2}
+)
 print(f"Result: {result}")
 
 # 6. Plain callable

{predikit-0.4.3 → predikit-0.4.5}/examples/02_xgboost_regression.py

@@ -1,7 +1,8 @@
 """
 Example 02 — XGBoost regression model wrapped as a tool.
-Requires: pip install modelbridge[xgboost]
+Requires: pip install predikit[xgboost]
 """
+
 import json
 
 from pydantic import BaseModel, Field
@@ -10,8 +11,8 @@ from sklearn.model_selection import train_test_split
 
 try:
     from xgboost import XGBRegressor
-except ImportError:
-    raise SystemExit("XGBoost not installed. Run: pip install modelbridge[xgboost]")
+except ImportError as err:
+    raise SystemExit("XGBoost not installed. Run: pip install predikit[xgboost]") from err
 
 from predikit import ModelTool
 
@@ -44,11 +45,13 @@ print("=== OpenAI Tool Schema ===")
 print(json.dumps(tool.to_openai(), indent=2))
 
 print("\n=== Invoke ===")
-result = tool.invoke({
-    "feature_1": float(X_test[0, 0]),
-    "feature_2": float(X_test[0, 1]),
-    "feature_3": float(X_test[0, 2]),
-    "feature_4": float(X_test[0, 3]),
-})
+result = tool.invoke(
+    {
+        "feature_1": float(X_test[0, 0]),
+        "feature_2": float(X_test[0, 1]),
+        "feature_3": float(X_test[0, 2]),
+        "feature_4": float(X_test[0, 3]),
+    }
+)
 print(f"Prediction: {result}")
 print(f"Actual:     {y_test[0]:.2f}")

{predikit-0.4.3 → predikit-0.4.5}/examples/03_orlando_real_estate.py

@@ -7,6 +7,7 @@ schema generation → registry export → direct invocation → callable.
 
 Requires: pip install xgboost
 """
+
 import json
 
 import numpy as np
@@ -16,8 +17,8 @@ from sklearn.model_selection import train_test_split
 
 try:
     from xgboost import XGBRegressor
-except ImportError:
-    raise SystemExit("XGBoost not installed. Run: pip install xgboost")
+except ImportError as err:
+    raise SystemExit("XGBoost not installed. Run: pip install xgboost") from err
 
 from predikit import ModelTool, ToolRegistry
 
@@ -27,31 +28,33 @@ from predikit import ModelTool, ToolRegistry
 rng = np.random.default_rng(42)
 n = 1_000
 
-sqft           = rng.integers(800, 4_000, n).astype(float)
-bedrooms       = rng.integers(1, 6, n).astype(float)
-bathrooms      = rng.choice([1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0], n)
-year_built     = rng.integers(1960, 2024, n).astype(float)
-has_pool       = rng.integers(0, 2, n).astype(float)
+sqft = rng.integers(800, 4_000, n).astype(float)
+bedrooms = rng.integers(1, 6, n).astype(float)
+bathrooms = rng.choice([1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0], n)
+year_built = rng.integers(1960, 2024, n).astype(float)
+has_pool = rng.integers(0, 2, n).astype(float)
 zip_code_group = rng.integers(0, 5, n).astype(float)  # 0=rural … 4=prime
 
 price = (
     80 * sqft
     + 15_000 * bedrooms
     + 10_000 * bathrooms
-    + 500   * (year_built - 1960)
+    + 500 * (year_built - 1960)
     + 20_000 * has_pool
     + 30_000 * zip_code_group
     + rng.normal(0, 15_000, n)
 ).clip(50_000, 1_500_000)
 
-df = pd.DataFrame({
-    "sqft":           sqft,
-    "bedrooms":       bedrooms,
-    "bathrooms":      bathrooms,
-    "year_built":     year_built,
-    "has_pool":       has_pool,
-    "zip_code_group": zip_code_group,
-})
+df = pd.DataFrame(
+    {
+        "sqft": sqft,
+        "bedrooms": bedrooms,
+        "bathrooms": bathrooms,
+        "year_built": year_built,
+        "has_pool": has_pool,
+        "zip_code_group": zip_code_group,
+    }
+)
 
 X_train, X_test, y_train, y_test = train_test_split(df, price, test_size=0.2, random_state=42)
 
@@ -61,17 +64,19 @@ X_train, X_test, y_train, y_test = train_test_split(df, price, test_size=0.2, ra
 model = XGBRegressor(n_estimators=200, learning_rate=0.05, random_state=42)
 model.fit(X_train, y_train)
 
+
 # ---------------------------------------------------------------------------
 # 3. Input schema — field names must match DataFrame columns exactly
 # ---------------------------------------------------------------------------
 class OrlandoHouseInput(BaseModel):
-    sqft:           float = Field(description="Total square footage of the home")
-    bedrooms:       float = Field(description="Number of bedrooms")
-    bathrooms:      float = Field(description="Number of bathrooms (0.5 increments)")
-    year_built:     float = Field(description="Year the home was built")
-    has_pool:       float = Field(description="1 if the home has a pool, else 0")
+    sqft: float = Field(description="Total square footage of the home")
+    bedrooms: float = Field(description="Number of bedrooms")
+    bathrooms: float = Field(description="Number of bathrooms (0.5 increments)")
+    year_built: float = Field(description="Year the home was built")
+    has_pool: float = Field(description="1 if the home has a pool, else 0")
     zip_code_group: float = Field(description="Area cluster 0–4 (0=rural, 4=prime location)")
 
+
 # ---------------------------------------------------------------------------
 # 4. Wrap as ModelTool and register
 # ---------------------------------------------------------------------------
@@ -99,11 +104,11 @@ print(json.dumps(registry.to_openai(), indent=2))
 # 6. Run a prediction
 # ---------------------------------------------------------------------------
 sample = {
-    "sqft":           2_200.0,
-    "bedrooms":       3.0,
-    "bathrooms":      2.0,
-    "year_built":     2005.0,
-    "has_pool":       1.0,
+    "sqft": 2_200.0,
+    "bedrooms": 3.0,
+    "bathrooms": 2.0,
+    "year_built": 2005.0,
+    "has_pool": 1.0,
     "zip_code_group": 3.0,
 }
 

{predikit-0.4.3 → predikit-0.4.5}/examples/04_confidence_routing.py

@@ -9,6 +9,7 @@ Demonstrates the three on_low_confidence modes:
 threshold=0.999 guarantees the low-confidence path fires for illustration.
 In production, set this between 0.75 and 0.95.
 """
+
 from pydantic import BaseModel, Field
 from sklearn.datasets import load_iris
 from sklearn.linear_model import LogisticRegression
@@ -26,9 +27,9 @@ fallback_clf = GaussianNB().fit(X, y)
 
 class IrisInput(BaseModel):
     sepal_length: float = Field(description="Sepal length in cm")
-    sepal_width:  float = Field(description="Sepal width in cm")
+    sepal_width: float = Field(description="Sepal width in cm")
     petal_length: float = Field(description="Petal length in cm")
-    petal_width:  float = Field(description="Petal width in cm")
+    petal_width: float = Field(description="Petal width in cm")
 
 
 fallback_tool = ModelTool(
@@ -61,7 +62,9 @@ print("=== Mode: warn ===")
 result = warn_tool.invoke(sample)
 print(result)
 if result.get("_low_confidence"):
-    print(f"  → confidence {result['_confidence']:.3f} below threshold; agent should seek clarification")
+    print(
+        f"  → confidence {result['_confidence']:.3f} below threshold; agent should seek clarification"
+    )
 
 # ---------------------------------------------------------------------------
 # Mode 2: raise — lets the calling agent handle uncertainty explicitly