Skip to content

Workflow

The workflow described in the diagram below outlines a structured process for managing data, modeling, and attribution creation in an analytical system. The workflow is divided into two key processes: Data Process and Modeling Process, each ensuring a streamlined approach to data ingestion, transformation, and model enhancement.

flowchart TB UD@{shape: docs, label: "Upload data
sources"} CDS[Create
Dataset] EDS[Extend
Dataset] EED{Extend
Existing?} ADS@{shape: lin-cyl, label: "Active
Dataset"} UM{Update
Models?} CA@{shape: processes, label: "Create
Attributions"} EEA{Extend
Existing?} EA@{shape: docs, label: "Extend
Attributionset"} CM@{shape: processes, label: "Refit Models"} CAS@{shape: doc, label: "Create
Attributionset"} subgraph dpr [Data process] direction LR UD --> |success| EED EED --> |no| CDS EED --> |yes| EDS CDS & EDS --> |activate| ADS end subgraph mpr [Modeling process] direction LR UM --> |no| CA --> EEA --> |yes| EA UM --> |yes| CM --> CA EEA --> |no| CAS end dpr --> mpr

Data Process

  1. **Upload Data Sources
    • The process begins with uploading relevant data sources, such as marketing, sales, customer engagement, or external factors.
  2. Extend Existing Dataset
    • If an existing dataset is available, and the uploaded data is a continuation of the excising data one should extend the excising data. If not one should create a new dataset.
  3. Activate Dataset
    • Once a dataset is created or extended, it becomes the Active Dataset, ensuring it is available for subsequent modeling and attribution tasks.

This Data Process ensures that datasets are efficiently managed, reducing redundancy while incorporating new information.

Modeling Process

  1. Update Models Decision
    • Assesses whether models need to be updated based on new or modified data.
    • If models do not require updating, one can proceed to create attributions.
    • If models do require updates, then one should preform a refit.
  2. Refit Models (If Needed)
    • If updates are required, models are refitted to accommodate the new dataset, ensuring improved accuracy and relevance.
  3. Create Attributions
    • Attribution analysis is conducted to measure the impact of different factors on key performance indicators.
  4. Extend Existing Attribution Set
    Decide if existing attribution set should be extended.
    • If yes, the attribution set is extended.
    • If no, a new attribution set is created.

This Modeling Process ensures that models are updated only when necessary, reducing computational overhead while maintaining analytical integrity.

Process Integration

  • The Data Process directly feeds into the Modeling Process, ensuring that models and attributions reflect the most up-to-date data.
  • This structured workflow optimizes efficiency by selectively updating models and extending datasets and attributions when required, maintaining a balance between computational efficiency and analytical rigor.

This workflow is designed to support scalable, automated, and data-driven decision-making, ensuring accurate predictions and actionable insights.