Maximizing VALUE from data
The Tenets Behind Daisho: Informed by lived experience across the entire Data Science pipeline
Data Science should be done as close to business operators as possible.
Data Scientists and Businesses speak different languages - translating takes up too much time. Translating loses too much information too. Value is maximized only when business users are enabled to do data science themselves.
Any product that intends to bridge the gap must speak the business language to the users.
Business problems need explainability (why, how) as much as accuracy.
Business problems always need actionability. This only comes from the ability to explain and justify any predictions being made or inferences being drawn.
Products aimed at business problems must, therefore, only use algorithms which lend themselves well to explainability and actionability.
Its actions that are the end-point of the analysis, not inferences.
People who do actions are different from those that do analysis. The tools for actions (e.g., SalesForce, SCADA, ZenDesk) are different too.
Hence, for maximal value addition, the product must be easy to integrate with other tools too.
DRY: Dont Repeat Yourself
More than 80% of a data scientist's work is either grunt-work, or repititive work. Doing all that manually kills productivity.
The product must support as much automation as possible, while retaining the flexibility to mix-and-match components as required.