Data analysis, in plain words
What the terms mean anywhere, and what they mean when you use Trinetro Labs, the world's first Automated Intelligence platform for data analysis.
- Automated Intelligence: A way of answering questions about data by reasoning through its real structure, step by step, instead of predicting a likely-sounding answer. The same question on the same data gives the same answer, and every step can be shown.
- Business metric: A named number a business runs on, such as revenue, margin, churn or average order value, defined once so that everyone who asks for it gets the same figure.
- Deterministic analysis: Analysis in which identical inputs always produce identical outputs. Asking the same question of the same data twice returns the same number, which is what makes a figure something you can audit.
- Fan-out (double counting): What happens when two tables are joined on a key that repeats: each row on one side is matched to several rows on the other, and a total taken after the join counts the same amount more than once.
- Governed team memory: A store of the definitions a team has agreed on: what "revenue" means, what a shorthand stands for. Personal definitions belong to one person; a company default is approved by an admin and can be withdrawn.
- Grain: What one row of a table stands for: one order, one order line, one customer per month. Knowing the grain is what tells you which numbers can be added together safely.
- Hallucination (in data analysis): An answer that sounds right but is not supported by the data: a column that does not exist, a join that was never declared, a number produced without a calculation behind it.
- Join path: The chain of tables and keys that connects two pieces of information, for example customers to orders to products. Several paths can exist, and they do not all give the same answer.
- Key Insights Engine: The part of a platform that looks at a result and points out what stands out in it, such as a dominant segment, a concentration, an imbalance, an outlier or a long tail.
- Natural language to SQL (NL2SQL): Turning a question written in everyday language into the database query that answers it, so that the person asking does not have to write SQL.
- Reasoning trace: The record of how an answer was reached: which table and columns were used, how tables were joined, what filters were applied, and why each choice was made.
- Semantic layer: A description of the data in business terms: which columns are measures, which are categories, how tables relate, and what each metric means. It sits between raw tables and the people asking questions.
- Text-to-SQL: A technique in which a language model is given a question and a description of the tables and asked to write a SQL query. It is flexible, and the query it writes can differ from one run to the next.
- Trinetro Execution Intelligence (TEI): The name of the reasoning engine inside Trinetro Labs, which users know as their Third Eye. It has six layers: Connector, Semantic, Planning, Execution, Intelligence and Memory.