Data Quality for Analytics Using SAS

Chapter 19: Influence of Data Correctness on Model Quality in Predictive Modeling

19.1 Introduction

General

Non-visible data quality problem

Random and systematic bias

Biased values in the scoring data partition

19.2 Simulation Methodology and Data Preparation

Standardization of numeric values

Inserting random biases in the input variables

Inserting systematic biases in the input variables

Inserting a random bias in the target variable

Inserting a systematic bias in the target variable

Simulation scenarios

19.3 Results for Random and Systematic Bias in the Input Variables

Scenario settings

Bias in the input variables in the training data only

Bias in the input variables in the training and scoring data

Comparison of results

19.4 Results for ...