Data Wrangling

Introduction

Each challenge hosted on the Spectral Platform has a starter training dataset that can be retrieved via Spectral CLI.

Spectral Python SDK provides Data Wrappers, which help facilitate fetching real-time data that can be used to participate in challenges (see Modeler Handbook's Step 7).

Details

• Fetch the training data through the Spectral SDK

• If required and deemed suitable, perform any data processing and/or additional feature engineering operations on top of the provided training dataset

• [Optional] The Modeler can use any other external data if they wish to. However, it is imperative that:

  • any data other than that provided by the Spectral SDK should be readily available in a timely manner during the Consumption Window

  • modelers retain all the necessary scripts pertaining to data fetching, feature engineering, feature selection, etc. as they will be required during Model Validation and Consumption phases