Evaluating statistics

One of the key property of Rulex Factory Data Manager task is the possibility to evaluate statistics on the underlying dataset through simply drag and drop operations and automatically populate a dedicated sheet visualization.

All these operations are performed in the Statistic Manager. General overview about this pane has been given in the sheets tab page.

Here we are going to concentrate on the general approach about statistic evaluation and to a more detailed description of the four areas composing the Statistic Manager.

Procedure

  1. On the Attribute list select a list of attributes and drag one of its chips onto the first areas of the Statistic Manager. Attribute chips are going to be shown in the dropped area selected.

  2. Default Single statistics are evaluated for the dropped attribute and upper rows of the current sheet are populated with the results

  3. Stats configuration can be modified through the dedicated statistics configuration panel available by clicking on the Pencil icon in the Statistics area.

  4. Take into account any of the presented operations can be undone and redone at any moment.

Input areas

Input areas correspond to the first two areas from the left to the right of the Statistic Manager. Here is where user provides input attributes for his evaluation. Any statistic evaluation requires at least an input attribute which is mandatory and provided by using the first area Var_1.

Some classes of statistics require further input information: those additional attributes are provided by using the second area of the Statistic Manager meaning the Var_2/Target column.

Any of these areas are organized in rows. Any row in any area can contain an arbitrary number of attributes. All these attributes will be used to construct the same unique table result. In case some configuration of a single row are modified the whole role will be re-evaluated and the corresponding table on the sheet erased and redrawn completely.

In case of a multivariate problem where two input attributes or an input attribute and a target attribute are used, if more than one input is contained in the two areas for a single row, the result constructed will be pairing the two list in the common indexes. If one of the two lists is longer, the remaining attributes are going to not be matched and therefore ignored.

You can move attributes from one area to another one or from one row to another one or simply re-ordering attributes inside the same row by using simple drag’n drop operations, similarly to what performed in attribute addition from the Attribute list.

Statistics area

In the Statistics area, we define the type of statistic evaluation the system has to compute for the current considered row.

All stats available in Rulex Factory are divided in two macro classes:

  • Univariate: which involves only one input attribute for its evaluation. The following subclasses own to this group:

    • Single statistics

    • Values, frequencies and quantiles

  • Bivariate: which involves two input attributes for its evaluation. The following subclasses own to this group:

    • Correlation/Covariance

    • Cross tabulation statistics

    • ROC Curve

    • Test for independent samples

    • Test for paired samples

Note

Form of the resulting table (size, color, styling, shape…) is defined by the selected subclass. Further configuration inside the same subclass only add extra information as extra columns or rows, extra values etc…

For each subclass, you can customize the list of stats to evaluate among the whole available list as well as some computation options. All these internal configurations are modified through a dedicated statistics configuration panel which differs for any subclass and which is explained in detail in this page.

The statistics configuration panel can be opened by clicking on the Pencil icon located next to the subclass statistics drop down menu in each row of the Statistics area.

Sheet area

In this area for each row you can configure the sheet where to display the evaluated results.

Only the Locked sheets are considered for automatic filling.

In each row we have a dropdown menu populated with the names of all the Locked sheets. You can lock an empty sheet at any time by clicking on the Lock icon located in the upper left corner of the considered sheet header tab.

Changing the sheet of a single row of the Statistic Manager will erase the associated result table from the previous sheet redrawing it in the new selected sheet.