Shared Utilities

Common libraries and global configuration management

The system architecture cleanly separates core ML logic (app-ml) from the UI layer (app-ui). However, both rely on a set of shared utilities and configurations to ensure consistency in paths, logging, and environment management.

​

Project Structure

The project root contains these shared assets:

​

Global Configuration

The config/config.yaml file acts as the single source of truth for infrastructure paths and service definitions that cross the ML/UI boundary.

config/config.yaml

# Used by both the Kedro pipeline and the Dash UI
data_manager:
  raw_data_path: "data/raw_data/raw_data_full.parquet"

ui:
  theme: "LUMEN"
  refresh_interval: 60

​

Common Library (common/)

This lightweight library prevents code duplication between the separate Docker services.

​

utils.read_config

A robust YAML reader used by the Dash app to load settings without hardcoding paths.

from common.utils import read_config
from pathlib import Path

# In app-ui/app.py
project_root = Path(__file__).parents[1]
config = read_config(project_root / 'config/config.yaml')

​

data_manager.py

Standardizes DataFrame loading to ensure the UI reads Parquet files exactly as the model produced them (handling timestamp parsing and category types consistently).