.project-metadata.yaml

name: Continuous Model Monitoring
description: >-
  Demonstration of how to perform continuous model monitoring on CML 
  using Model Metrics and Evidently.ai dashboards
author: Cloudera Inc.
specification_version: 1.0
prototype_version: 2.0
date: "2022-04-01"

environment_variables:
  DEV_MODE:
    default: False
    description: >-
      Flag to indicate if the AMP should run on a 5% sample of the dataset 
      (True) to facilitate efficient project development or the full dataset (False).
    required: True

feature_dependencies:
  - model_metrics

runtimes:
  - editor: PBJ Workbench
    kernel: Python 3.9
    edition: Standard

tasks:
  - type: run_session
    name: Install Dependencies
    script: scripts/install_dependencies.py
    memory: 2
    cpu: 1

  - type: run_session
    name: Prepare Data
    script: scripts/prepare_data.py
    memory: 2
    cpu: 1

  - type: run_session
    name: Train Model
    script: scripts/train.py
    memory: 4
    cpu: 2

  - type: create_model
    name: Price Regressor
    entity_label: pr_model
    description: Predict housing prices with ridge regression
    access_key_environment_variable: SHTM_ACCESS_KEY
    default_resources:
      cpu: 1
      memory: 2
    default_replication_policy:
      type: fixed
      num_replicas: 1

  - type: build_model
    name: Price Regressor
    entity_label: pr_model
    target_file_path: scripts/predict.py
    target_function_name: predict
    examples:
      - request:
          {
            "record":
              {
                "id": 1962200037,
                "price": 626000,
                "bedrooms": 3,
                "bathrooms": 2.25,
                "sqft_living": 1750,
                "sqft_lot": 1572,
                "floors": 2.5,
                "waterfront": 0,
                "view": 0,
                "condition": 3,
                "grade": 9,
                "sqft_above": 1470,
                "sqft_basement": 280,
                "yr_built": 2005,
                "yr_renovated": 0,
                "zipcode": 98102,
                "lat": 47.6498,
                "long": -122.321,
                "sqft_living15": 2410,
                "sqft_lot15": 3050,
                "date_sold": "2014-05-02",
                "date_listed": "2014-03-04",
              },
          }

  - type: deploy_model
    name: Price Regressor
    entity_label: pr_model

  - type: run_session
    name: Run Simulation
    script: scripts/simulate.py
    memory: 4
    cpu: 2