|
|
|
|
|
|
|
|
|
Retrospective data analysis
| Глоссарий по искусственному интеллекту |
A modeling process that is designed to understand past trends or events. a common problem in data analysis or data mining is that the objective is to predict future events based on past events, but that data are all based on past events. a common solution is to cross-validate the analysis by using sampling, "hold one out" or bootstrap analyses. each of these techniques attempts to provide "new" data by only using part of the database by only using part of the data to form the data and then using the remainder to validate the model. see also: cross-validation, bootstrapping.
|
|
Understand, английский
Validation, английский
- Confirmation that particular requirements for a specific intended use are fulfilled.
- Action or process of proving that a procedure, process, system, equipment, or method used works as expected and achieves the intended results. [clsi]
- Утверждение
- Подтверждение соответствия; проверка
- Валидация. подтверждение путем экспертизы и представления объек-тивного доказательства того, что особые требования, предназначенные для конкрет-ного применения, соблюдены [4]. см. также method validation (валидация метода).
- The process of determining the correctness of a standard as to its technical completeness and lack of ambiguity. (ieee)
- The process in which microsoft tests an app to make sure it meets app requirements.
- The process of comparing files on local volumes with their associated data in secondary storage by remote storage. volumes that are validated ensure that the correct data is recalled from remote storage when a user attempts to open the file from a local volume.
- The process of confirming that data passing into the system is correct and complies with predefined rules, definitions, or parameters.
- Валидация. подтверждение путем экспертизы и представления объективного доказательства того, что особые требования, предназначенные для конкретного применения, соблюдены [4]. см. также method validation (валидация метода).
- Авторизация
Bootstrapping, английский
- Bootstrapping can be used as a means to estimate the error of a modeling technique, and can be considered a generalization of cross-validation. basically, each bootstrap sample from the training data for a model is a sample, with replacement from the entire training sample. a model is trained for each sample and its error can be estimated from the unselected data in that sample. typically, a large number of samples (>100) are selected and fit. the technique has been extensively studied in statistics literature.
- Creating a theoretical spot rate curve using one yield projection as the basis for the yield of the next maturity. bootstrapping follows the work of efron. it involves a monte carlo approach.
- A technique for constructing standard errors and conducting hypothesis tests that requires no distributional assumptions and works by resampling from the data.
|
Random world model, английский
A model for reasoning that assumes all basic events are equally likely. also known as the "principle of insufficient reasoning." the underlying assumption behind rough set theory. see also: rough set theory.
Additive models, английский
A modeling technique that uses weighted linear sums of the possibly transformed input variables to predict the output variable, but does not include terms such as cross-products which depend on more than a single predictor variables. additive models are used in a number of machine learning systems, such as boosting, and in generalized additive models (gams). see also: boosting, generalized additive models. additivity and variance stabilization (avas) avas, an acronym for additivity and variance stabilization, is an modification of the ace technique for smooth regression models. it adds a variance stabilizing transform into the ace technique and thus eliminates many of ace`s difficulty in estimating a smooth relationship. see also: ace.
|
|
|
|
|
|
|
