Feedback Request
FINDHR
- Data - Population: when biases are produced by the data, the most classic example is labelling errors.
- Population - Sample: when the population is not well represented in the sample (e.g. population bias, survey bias, seasonal bias, etc.).
- Sample - Variables and values: when the data of the sample is not well represented by the values or the variables (e.g. oversimplification problem, omitted variable, etc.).
- Variables and values - Patterns: when the data used can mislead patterns or assumptions (e.g. measurement bias, over and underfitting, etc.)
- Patterns - Predictions: when the resulting patterns of the analysis suggest outcomes that can be biased or incorrectly adapted to the real world (e.g. aggregation bias, model selection problems, etc.).
- Predictions - Decisions: when the analysis obtained mislead decisions in the real world (e.g. visualization errors).
Decisions - World: when the decisions made by the algorithm affect the real-world application and use of the algorithm (e.g. automation bias, accessibility, etc.).
Ein Problem melden
Ist dieser Inhalt unangemessen?
0 Kommentare
Kommentare werden geladen ...
Kommentar hinzufügen
Melden Sie sich mit Ihrem Konto an oder registrieren Sie sich, um Ihren Kommentar hinzuzufügen.
Kommentare werden geladen ...