matrice-analytics
raw JSON → 0.1.234 verified Mon Apr 27 auth: no python
Post-processing analytics for Matrice.ai inference pipelines. Currently at version 0.1.234, released with no fixed schedule.
pip install matrice-analytics Common errors
error ImportError: cannot import name 'AnalyticsPipeline' from 'matrice_analytics' ↓
cause Old import path or outdated package version.
fix
Upgrade to latest version (pip install --upgrade matrice-analytics) and use 'from matrice_analytics import AnalyticsPipeline'.
error matrice_analytics.exceptions.AuthenticationError: API key not provided ↓
cause Environment variable MATRICE_API_KEY not set or not passed to constructor.
fix
Set MATRICE_API_KEY environment variable or pass api_key='your_key' to AnalyticsPipeline().
error AttributeError: 'Result' object has no attribute 'items' ↓
cause Code expects dict but got Result object after upgrade to >=0.1.200.
fix
Use result.summary() or access attributes like result.accuracy instead of iterating over items.
Warnings
gotcha The library expects an API key via environment variable 'MATRICE_API_KEY' or constructor argument. Missing key leads to runtime authentication error. ↓
fix Set MATRICE_API_KEY environment variable or pass api_key to constructor.
breaking Version 0.1.200 changed the default output format from dict to custom Result object. Old code relying on dict access will break. ↓
fix Use .summary() method or access attributes directly on Result object instead of dict keys.
deprecated The function 'quick_analyze' is deprecated since version 0.1.150. Use 'AnalyticsPipeline.run()' instead. ↓
fix Replace calls to quick_analyze with pipeline.run().
Imports
- AnalyticsPipeline wrong
from matrice_analytics.pipeline import AnalyticsPipelinecorrectfrom matrice_analytics import AnalyticsPipeline
Quickstart
from matrice_analytics import AnalyticsPipeline
import os
pipeline = AnalyticsPipeline(api_key=os.environ.get('MATRICE_API_KEY', ''))
result = pipeline.run("sample_data")
print(result.summary())