Skip to content

Latest commit

 

History

History

11. Sagemaker

Sagemaker-pipeline-1 (basic pipeline)

  • environment pipeline
  • data preparation
  • modelling
  • deployement
  • evaluation
  • prediction

Environment Setup

  1. get aws role
  2. region and other aws configuational artifacts

Data Preparation

upload the raw data to S3 bucket
sagemaker conatiner based training read data only from S3

Modelling

train sagemaker based xgboost model

Deployement

Deploy the xgboost model as an endpoint Mention:

  • instance_type : 'ml.m4.xlarge'
  • initial_instance_count : 1 After this step we have successfully deployed ML model as an Endpoint

Evaluation

  1. generate confusion martrix
  2. Pr curve

Prediction

Predict using endpoint