AI Dictionary › AI Fundamentals
Valori mancanti (Missing Values)
Missing values are the empty or absent cells in a dataset, points where an observation has no recorded data for a particular feature: a customer who did not report their age, a sensor that stopped working for a few minutes, a field left blank on a form. They are one of the most common and insidious problems in real-world machine learning practice, because most algorithms do not know how to handle them directly.
Before even deciding how to treat them, it is important to understand why they are missing: a value can be missing completely at random, or its absence can depend on other observed variables, or even on the value itself that is missing, for example people with very high incomes who tend not to report it. The most common handling strategies are removing rows or columns with too many missing values, when they are few and not systematic, and imputation, replacing absent values with plausible estimates, ranging from the simple mean or median of the variable to more sophisticated predictive models that estimate the missing value based on other available features.
In applied AI, handling missing values is an almost mandatory step before training: many algorithms, like neural networks and support vector machines, simply fail if given incomplete data, while some tree-based models, like certain gradient boosting implementations, can handle them natively. It is a particularly delicate issue in healthcare, where a missing data point may hide a relevant clinical pattern, and in surveys, where the non-response rate always needs careful assessment.
The systematic study of missing data belongs to late twentieth-century statistics, with a foundational contribution from the work of Donald Rubin, who in 1976 formalized the distinction between the different mechanisms by which data can be missing, a classification still at the basis of how statisticians and data scientists reason about the problem today. With the spread of machine learning on real, often incomplete data collected under uncontrolled conditions, handling missing values became an essential practical skill for anyone working with data.
From our network
INDACO TMS: Transport Management for European Logistics
Shipment tracking, multi-carrier EDI and automated invoicing in one cloud platform. Invoices generated in under 10 seconds.
Visit indacotms.com →From the Agora Intelligence blog
📱 Download the Android app (beta) iOS coming soon
Say what you mean. Get what you need.
Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.