Any individual/family property (for details on property codes see here) can be used to split he corpus into subcorpuses. A subcorpus (whose title indicates the parent corpus as well as the property label and value) has the appearance of an autonomous corpus (with its own corpus window and all dependant windows), and every operation on a corpus can also be effectuated on a subcorpus. The important difference is that a subcorpus remains linked to the parent corpus, and all relation and circuit search processes, while limiting results to individuals in the subcorpus, always run through the total corpus.
The partitioning commands are located on a specific bar, placed on the top of the PUCK Main Window.
Warning : a subcorpus that is saved and re-imported as a normal corpus loses this important subcorpus property. All links to individuals outside the subcorpus are cut, and external individuals can no longer act as intermediaries for chains between members of the subcorpus.
To create a partition, first click on the Add Segment button (symbol "+") on the Partitions Bar, so that the Partition Criteria Input window will open (see snapshot below).
Here, the Model drop-down menu enables to choose between Individual or Family, which are the two models of partition criteria. The Label drop-down menu contains a list of endogenous and exogenous properties, that can be set as partitioning criteria. Note that PUCK automatically integrates into the list the attributes/additional data that you might have defined during the encoding phase.
For an explanatory list of the properties codes, see here.
The Parameter field allows using as partitioning criterion some properties, such as "PEDG" or "PROG", that require a specification to operate.
For example, to segment the dataset on the basis of the number of known ascendants of a given degree, the Model field must be set to "INDIVUDUAL" and the Label field to "PEDG". In the Parameter field, you will then define the maximal generational depth to which calculate the individuals pedigree. If you set the Parameter at "2" and the Type at "Raw", PUCK will create five clusters :
- "0" (composed by individuals whose grand-parents are all unknown) ;
- "1" (one grand-parent known) ;
- "2" (two grand-parents known) ;
- "3" (three grand-parents known) ;
- "4" (all grand-parents known).
The checkbox list called Type allows choosing, between several possibilities, the most pertinent way to regroup the clusters of a given partition. This choice depends on the partitioning criterion. Here is a brief description of how these operators work :
- Raw : produces a basic partition containing as much clusters as necessary. E.g. the Gender partitioning (Model : "INDIVIDUAL", Label : "GENDER") produces 3 clusters : a Null one, a Female and a Male ones. A last name partition (Model : "INDIVIDUAL", Label : "LASTN") as many clusters as the number of last names in the corpus (plus, eventually, a null cluster) ;
- Binarization : always produces 2 clusters, one corresponding to the Pattern and the other regrouping everything but the Pattern ;
- Free grouping : allows fixing irregular (or regular) date intervals ;
- Counted grouping : allows fixing the number of clusters in which you want to divide a given period (from start … to end) ;
- Sized grouping : allows fixing the duration of clusters in which you want to divide a given period (from start … to end).
Note : The dataset can be partitioned more than once, so you can superimpose different partitioning criteria. This can be done by running once again the Add Segment button (symbol "+"). So, the successive application of different properties as partition criteria permits to refine partitioning which can be necessary for your analysis process.
***GAP Family Scope
The Family Scope checkbox list allows [...]
Navigate through partitions and clusters
After partitioning the dataset, you can navigate through the different partitions and clusters, by using the Partition / Cluster drop-down menus, or by clicking on the Up / Down one segment buttons, placed on the right side of the instruments bar ("Δ"/"∇" symbols on the bar). All the individuals and families appearing in the Individuals and Families navigation tabs, belong then to the selected cluster. However, Ego's relatives don't necessarily find themselves in his same cluster ; and the selected cluster changes if you double-click on an individual who doesn't belong to the previous cluster. So be careful : if you navigate through the corpus by clicking on individuals, you could jump out of the starting cluster without knowing it!
When necessary, you can manage partitions by selecting one of them and giving the following commands :
- Remove current segment (the "-" symbol on the bar), which erase the selected partition ;
- Clear all segments (the "broom" symbol on the bar), which erase all partitions.
Warning : partitions are hierarchically organized. If you erase the first partition (see the Partition drop-down menu) by using the command "-", the action will be effective on all other partitions too.
- Genealogical corpora can be partitioned on a first level according to Gender (Model: INDIVIDUAL, Label: SEX) and then, on a second level, one of the clusters can be partitioned to the Family Name (Model: INDIVIDUAL, Label: LASTN) and one of these clusters is to be partitioned according to the surname (Model : INDIVIDUAL, Label FIRSTN). Use the arrow buttons to move up and down to the different levels of partition. For each item from the « Partition » list, the correspondent clusters appear in the « Cluster » list (Fig. 26).
- Partitions the corpus according to the date of birth of individuals: Start = 1500 ; Size = 100 and End = 1900 divides individuals born in 1500 to 1900 onwards into sub-sets corresponding to 100 year intervals, adding another sub-set containing individuals born before 1500 and a still further sub-set containing individuals whose birth date is unknown. (Sized grouping) (Fig. 27)