Quality — Purity, Accuracy and Fidelity of Data in the Tabular Model

1. Definition — What QUALITY means in Power BI

In Power BI, Quality is the structural property that measures how faithfully data represents the reality it claims to describe.

Data is high-quality when it is:

• Accurate (reflects measured reality)
• Precise (stable in repetition)
• Complete (not missing necessary components)
• Pure (free from corruption or noise)
• Valid (within the allowed domain)
• Coherent (harmonized with the rest of the model)

This property is evaluated across the entire pipeline:

• Data sources
• Power Query (extraction, transformation, cleansing)
• Tabular Model
• Relationships
• DAX measures
• Visualizations
• Refresh pipelines

Quality answers the fundamental question:
Can I trust this data before analyzing it?

2. Nature — Why QUALITY exists as a structural property

Data quality emerges from the interaction of three components:

A) The reality the data intends to represent

Example: sales amount, transaction date, customer status.

B) The capture and transformation process

Each step introduces risk of distortion:

• extraction
• manual entry
• legacy systems
• integration
• manual transformation

C) The semantic business model

Data only has quality if it fits correctly within the business meaning.
Quality is ontological: if the representation is not faithful, the data does not exist as useful information.

In Power BI, the tabular engine cannot “guess” business intent; it depends entirely on the quality of what it receives.

3. Function — How QUALITY operates in the Power BI pipeline

A) In Power Query (Data shaping)

Quality ensures:

• correct formats
• valid values
• well-defined domains
• correct imputations
• removal of invalid duplicates
• identity standardization

B) In the Tabular Model

Quality determines:

• key integrity
• relational stability
• accuracy of aggregations
• correct filter context behavior

C) In DAX

Quality defines:

• truthfulness of KPIs
• robustness of measures
• stability under slicers, drill-downs, and complex filters

D) In Visualizations

Quality guarantees:

• undistorted charts
• consistent numbers
• trustworthy interpretations

4. Consequences — What happens when Quality fails

A quality failure produces highly visible effects:

A) Incorrect KPIs

Sales, revenue, profit or volume metrics do not reflect reality.

B) Wrong conclusions

The business model can be interpreted incorrectly.

C) Distorted forecasting

Poor quality affects:

• forecasts
• predictive models
• correlations
• trend analysis

D) Broken relationships

Incorrect keys → no matching → blank or duplicated results.

E) Impossible DAX outputs

SUM, AVERAGE, MIN/MAX, COUNT return values misaligned with reality.

F) Toxic outliers

Extreme erroneous values ruin distributions and averages.

5. Interactions — Properties related to QUALITY

• Consistency
• Validity
• Integrity
• Precision / Accuracy
• Semantic Congruence
• Freshness

6. Evaluation Methods — How to measure QUALITY in Power BI

A) Data Profiling (Power Query)

• column quality
• distribution
• uniqueness
• null values
• anomalous values

B) Domain Validation

• valid dates
• expected categories
• plausible numeric values
• allowed ranges

C) Statistical Quality Tests

• normal vs real distribution
• outliers
• standard deviation
• dispersion tests

D) Business Logic Validation

• source systems
• business rules
• user expectations

E) Truth Tables

Comparison between:

• expected
• observed
• calculated

7. Applicable Models — Frameworks to ensure or restore QUALITY

A) Data Quality Scorecard

• accuracy
• completeness
• validity
• consistency
• uniqueness

B) Dimensional Quality Framework (DQF)

Organizes quality by entities, dimensions, facts and key attributes.

C) Tabular Data Health Matrix

Evaluates each table based on null %, duplicates %, validity and historical stability.

D) Valid Value Domain Mapping

Defines valid domains and dictionaries for coherent classifications.

E) Column Quality Classifier

Classifies each column by purity, stability, type and structural integrity.

8. Samples — Implementation in Power Query and DAX

A) POWER QUERY — Remove nulls

#"Remove Null Sales" =
Table.SelectRows(Sales, each [Amount] <> null and [Amount] > 0)

B) POWER QUERY — Validate category domain

#"Valid Category" =
Table.SelectRows(Products, each List.Contains({"A","B","C"}, [Category]))

C) DAX — Anomaly flag

Sales_Anomaly =
IF([Total Sales] > 1000000, 1, 0)

D) DAX — Expected range validation

ValidPrice =
IF(
    AND(
        MIN(Products[Price]) >= 0,
        MAX(Products[Price]) <= 5000
    ),
1,0)

E) DAX — Truth Table

EVALUATE
ROW(
   "Expected", [Expected Sales],
   "Actual", [Total Sales],
   "Quality_OK", IF([Expected Sales] = [Total Sales], TRUE(), FALSE())
)