AWS AIF-C01 ML Developing

Machine Learning Development Lifecycle

Business goal identification

Defining the success criteria or key performance indicators (KPIs) for the ML workload is critical.
為 ML 工作負載定義成功標準或關鍵績效指標 （KPI） 至關重要

ML problem framing ML

converting it into a machine learning problem 轉換為機器學習問題。

Data processing

Amazon SageMaker Data Wrangler

low-code no-code (LCNC) tool, mport, prepare, transform, featurize, and analyze data by using a web interface. Customers can add their own Python scripts and transformations to customize workflows.

低代碼無代碼 （LCNC） 工具。它提供了一個端到端解決方案，用於使用 Web 介面導入、準備、轉換、特徵化和分析數據。客戶可以添加自己的 Python 腳本和轉換來自定義工作流。

Amazon SageMaker Studio Classic

built-in integration of Amazon EMR and AWS Glue interactive sessions to handle large-scale interactive data preparation and machine learning workflows within your SageMaker Studio Classic notebook.
內置了 Amazon EMR 和 AWS Glue 互動式會話的集成 ，可在 SageMaker Studio Classic 筆記本中處理大規模互動式數據準備和機器學習工作流。

SageMaker Processing API

can run scripts and notebooks to process, transform, and analyze datasets ]

Managing Features

Amazon SageMaker Feature Store

Features stored in the store can be retrieved and enriched before being served to the ML models for inferences.

存儲中的特徵可以在提供給 ML 模型進行推理之前進行檢索

Model training and evaluation

Amazon SageMaker

• Customers aiming at a LCNC option can use Amazon SageMaker Canvas.
  以 LCNC 選項為目標的客戶可以使用
• saves the resulting model artifacts in an Amazon Simple Storage Service (Amazon S3) bucket that can be used later for inference.
  生成的模型構件保存在 Amazon Simple Storage Service （Amazon S3） 儲存桶中，以便稍後用於推理。

Amazon SageMaker JumpStart

pretrained, open source models that customers can use for a wide range of problem types.

預訓練的開源模型，客戶可以將其用於各種問題類型。

Model evaluation

Model evaluation datasets

The data you use is partitioned into three parts: training set, validation set, and test set
數據分為三個部分：訓練集、驗證集和測試集。訓練集用於訓練模型

Test set

one last time to ensure its predictive quality meets your standards.
進行最後一次測試，以確保其預測品質符合您的標準。

Classification and regression problems

Classification metrics

• Accuracy 準確性
• Precision 精度
• Recall 召回
• F1
• AUC-ROC
  
  ROC is a probability curve, and AUC represents the degree or measure of separability.
  ROC 是概率曲線，AUC 表示可分離性的程度或度量。

Regression metrics

• Mean squared error 均方誤差
  
  The smaller the MSE, the better the model’s predictive accuracy.
• R squared R 平方
  R squared explains the fraction of variance accounted for by the model.
  R 平方 （R squared） 解釋模型所佔的方差比例。

Amazon SageMaker Experiments

experiment with multiple combinations of data, algorithms, and parameters, all while observing the impact of incremental changes on model accuracy.

試驗數據、演算法和參數的多種組合，同時觀察增量更改對模型準確性的影響。

Amazon SageMaker Automatic Model Tuning

Hyperparameter tuning is a way to find the best version of your models.

超參數優化是查找模型最佳版本的一種方法。

Deployment

Monitoring

Amazon SageMaker Model Monitor

observe the quality of SageMaker ML models in production, detecting violations of user-defined thresholds for data quality, model quality, bias drift, and feature attribution drift.

客戶可以觀察生產中 SageMaker ML 模型的品質, 檢測違反使用者定義的數據品質、模型品質、偏差漂移和特徵歸因漂移閾值的行為，説明保持模型品質。