International Business Machines Corporation

Martin Keller

Calisto Paul Zuzarte

Dieter Helmut Roller

Andreas Arning

Christoph Lingenfelder

Stefan Raspl

Jayanta Basak

Laurent Sebastien Mignet

Gregor Meyer

Sourashis Roy

The proposed prediction of trends is based on history values of an observable. In more detail, a mathematical model is fitted on history values of an observable, said mathematical model having a ratio factor defined as a ratio between a number of history values used for calculation of a given observable value and a number of observable values modelled simultaneously. Information defining a minimum forecast horizon and a maximum forecast horizon together with respective ratio factor values is received, together with information defining a desired forecast horizon. A desired ratio factor value corresponding to the desired forecast horizon is determined based at least on said desired forecast horizon and said minimum and maximum forecast horizons together with said respective ratio factor values. The desired ratio factor value is used in fitting said mathematical model on said history values of said observable.

Prediction of trends based on history values of an observable

A window size for outlier detection in a time series of a database system is determined. Strength values are calculated for data points using a set of window sizes, resulting at least in one set of strength values for each window size. The strength values increase as a distance between a value of a respective data point and a local mean value increases. For each set of strength values, a weighted sum is calculated based on the respective set of strength values. A weighting function is used to suppress the effect of largest strength values and a window size is selected based on the weighted sums.

Determining a window size for outlier detection

A system and method for accessing a relational database and estimating the selectivity of a query (e.g., an SQL query) in order to better predict the number of qualifying records for simple and complex queries. A dataset is created based upon queries applied against the database by the user community. The dataset is populated with information related to query conditions and their respective combinations. A regression function reflecting correlations between query conditions is generated and used as a data mining model to calculate table-specific estimates for the cardinality of subsequent queries. An appropriate access method is then selected from a set of available access methods based upon the number of estimated query-qualifying records. By periodically updating the regression model with FIFO managed queries, a self-tuning mechanism is achieved resulting in better selectivity/result size estimates for use in selecting access methods used in compiling subsequent SQL queries applied against the database.

Self tuning database retrieval optimization using regression functions

A system and associated method for data mining prediction is presented according to which the user selects a database table by means of a graphical user interface. Some records in the table are complete, while other records are incomplete. A subset of records of the database table is determined wherein each record of the subset contains a data value in the column selected for prediction. This subset of records is used to generate a model by means of a data mining algorithm, such as linear regression, radial basis function, decision tree or neural network methods. The resulting model is then utilized to predict the empty data fields in the column. After completing the prediction, the predicted values are entered into the column for display to the user.

System and method of using data mining prediction methodology

A computerized data mining method for determining a prediction model for a dependent data mining variable based on a multitude of independent data mining variables, comprising replacement of the independent data mining variable with potential values from a global range by a multitude of independent local data mining variables, each relating to potential values from a sub-range of the global range. The method further comprises an initialization step and a looping sequence that includes steps of determining for every independent local data mining variable not yet reflected in the current prediction model a multitude of partial regression functions, determining for each of the partial regression functions a significance value, selecting the most significant partial regression function and the corresponding not yet reflected local data mining variable, and adding the most significant partial regression function to the current prediction model.

Partial stepwise regression for data mining

A technique for determination of an exception in multi-dimensional data is provided. The method determines an expected value for a set of cells of the multi-dimensional data by means of an ANOVA based method. A residual for each cell of the set of cells is then determined. Scaling of the residuals is then carried out, and the scaled residuals are then compared with a threshold value for determination of the exception.

Technique for determination of an exception in multi-dimensional data

Means and a computerized method for determining variable dimensions for a multi-dimensional database table, whereby the database table also comprises at least one pre-defined measure dimension for storing values of one or more measure types. Input records are treated as an implicit, yet unknown functional relationship between the measure types as dependent variables and the variable types as independent variables. A regression function is then calculated and used as a prediction model for the measure types based on the variable types, using the input records. The most significant variable types contributing to this prediction model are selected as variable dimensions for the database table.

Growing community of inventors

Vaihingen, Germany

Martin Keller