The power of the FSDA toolbox is easily deductible by taking a look at the introductory video on the toolbox features. Everything shown in the video, as in the following ones, is extensively documented in the toolbox guide.

FSDA TOOLBOX OUTLOOK

You can have a look at the toolbox features through the short following video or through the following didactic movies.

Movie 1: FSDA TOOLBOX applied to Hawkins dataset.

This movie shows the close correspondence betweeen two forward plots: the scaled residuals and the minimum deletion residuals among observations not in the subset. By looking at each of them, it is obvious that there are three groups of outliers. By brushing the three groups, once at a time, in the scaled residual plot, we realise that the groups correspond to the three peaks in the final part of the minimum deletion residual plot.

To find out more about this dataset please see Chapter 2 of "Robust Diagnostic Regression Analysis".

Movie 2: FSDA TOOLBOX applied to Fishery dataset.

This movie shows that linking different forward plots can help in finding unexpected patterns in real complex data. The example is about international trade data characterised by many informative variables. Here the forward search is used to analyse the value and the quantity traded among the European Union and a specific third country.

To find out more about this dataset please see Perrotta and Torti (2009): Detecting price outliers in European trade data with the forward search. In Data Analysis and Classifcation: From Exploration to Confirmation, Studies in Classifcation, Data Analysis, and Knowledge Organization. Springer, Berlin-Heidelberg. Forecoming.

Movie 3: FSDA TOOLBOX applied to AR dataset

This movie shows how masking effects can be easily detectd by looking at forward plots. The movie also presents the effect of the outliers on the significance of one variable throught the so called added variable plot with the regression line coming from the addition of that variable when the potential outliers are included or excluded. The dataset was introduced by Atkinson and Riani in 2000. It contains 60 observations on a response with the values of three explanatory variables.

To find out more about this dataset please see Chapter 2 of "Robust Diagnostic Regression Analysis".

Movie 4: FSDA TOOLBOX applied to Loyalty cards dataset

This movie shows how the forward search can be used to decide whether and how to transform the data. The data in the example consist of 509 observations on the behaviour of customers with loyalty cards from a supermarket chain in Northern Italy. The response (y) is the amount in euros spent at the shop over six months and the explanatory variables are: X1, the number of visits to the supermarket in the six month period; X2, the age of the customer and, X3, the number of members of the customer's family.

To find out more about this dataset please see Atkinson and Riani (2006), JCGS.

Movie 5: FSDA TOOLBOX applied to Hospital dataset

This movie shows that linking different forward plots can help in finding unexpected patterns in real complex data. The example shows the the first 54 observations coming from the first hospital cannot be considered homogeneous with respect to the last 54 observations coming from the second hospital.

To find out more about this dataset please see Riani and Atkinson (2007): Fast calibrations of the forward search for testing multiple outliers in regression. ADVANCES IN DATA ANALYSIS AND CLASSIFICATION. vol. 1, pp. 123-141.

FSDA Toolbox Download