All topics
21 / 55
Analyticsbeginner

Building Analyses: Criteria, Layout & Views

Explain the core workflow of building an OBIEE analysis — selecting criteria columns, adding filters, and composing views for presentation.

An analysis is the basic reporting object. Dashboards, agents and action links are all built on top of them.

The Criteria tab is like choosing your ingredients and recipe once; the Results tab is like deciding whether to plate that same dish as a formal course, a buffet spread, or a photo for the menu — same food, different presentations.

Key Concepts

1
Two tabs do the work.
    Criteria   what data: subject area, columns, filters, sorting
    Results    how it looks: tables, pivots, charts, gauges
2
Criteria first. Drag presentation columns in, add filters, and optionally write a column formula for one-off logic:
3
    CASE WHEN Revenue > 100000 THEN 'Large'
         WHEN Revenue > 10000  THEN 'Medium'
         ELSE 'Small' END
4
Where a calculation belongs is a real decision.
    in the analysis   one report needs it. Quick, and invisible
                      to everyone else.
    in the RPD        several reports need it, or it is a business
                      definition -- margin, active customer
5
A margin calculation written into fifteen analyses is fifteen definitions that will diverge. That belongs in the RPD, where there is one.
6
One result set, many views. The Results tab composes views over the same data:
7
    Title, Table, Pivot Table, Graph, Gauge, Narrative,
    Column Selector, View Selector, Legend, Static Text
8
They share one query. Adding a chart beside the table does not run the query again — both render the same result set. That is why a table and three charts on one analysis is cheaper than four analyses.
9
The compound layout arranges them. A chart above a table, both responding to the same filters, is the standard pattern.
10
Filters against selection steps, which is a distinction worth knowing.
11
    filter          applied in the query -- the database returns
                    fewer rows
    selection step  applied after -- the rows come back and are
                    then narrowed
12
Filter for anything that can be filtered. Selection steps are for member-level operations the query cannot express.
13
Column selectors and view selectors let the user change what is shown without a new analysis — swap the measure, or switch between a chart and a table. One object instead of six.
14
The habit worth having. Check the generated logical SQL from the Advanced tab. It shows exactly what was requested, which is how you confirm a filter is being applied in the query rather than after it.
15
What the interviewer is probing.1. "What is the difference between criteria and results?" Probing: the basics. Stalls: "Nothing." Moves up: criteria defines the columns, filters and selections; results holds the views that present them.
16
2. "What is a compound layout?" Probing: the vocabulary. Stalls: "The page." Moves up: the arrangement of several views — table, chart, title — that make up one analysis's output.
17
3. "Why avoid returning more rows than you display?" Probing: the performance point. Stalls: "No reason." Moves up: everything is fetched and held in memory even if a view shows a fraction, so filtering belongs in the criteria.
18
4. "What is a selection step?" Probing: the lesser-known feature. Stalls: "A filter." *Moves up:* post-query member manipulation such as keep-only, add or sort by hierarchy, applied after the result set is returned.
Use cases:analysis-buildingviewsreporting-fundamentals