---
myst:
    substitutions:
        bloc1:
            python: 1-5,16-60,65-76,81-121
            cpp: 3-6,12-41,46-51,56-96,105
        bloc2:
            python: 83-98
            cpp: 61-71
        bloc3:
            python: 110-115
            cpp: 85-90
        bloc4:
            python: 117-121
            cpp: 92-96
        bloc5:
            python: 13-74
            cpp: 16-77
---
(use_cases_macro_dataserver_pca_example)=
# Macro "**dataserverPCAExample.{{extension}}**"

## Objective

The goal of this macro is to show how to handle a PCA analysis. It is not much discussed here, as a 
large description of both methods and concepts is done in [](#dataserver_pca).

## Macro {{uranie}}

{{
    "```{literalinclude} " + parent_dir
    + "/roottest/uranie/doc/dataserver/use_cases/" + language + "/dataserverPCAExample." 
    + extension 
    + "\n"
    + ":language: " + language + "\n"
    + ":lines: " + bloc1[language] + "\n"
    + "```"
}}

This first part of this macro is described in [](#dataserver_pca). We will focus here on the second 
part, where all the numerical results are dumped on screen. These results are stored in the {{ds}} 
for the points and in a dedicated ntuple for the variable that can be retrieved by calling the method 
`getVariableResultNtupleD`. In both cases, the results can be split into two kinds:

- the quality of the representation: it is called *"cosca_X"* as it is a squared cosinus of the 
projection of the source under study (point or subject) on the X-th PC.
          
- the contribution to axis: it is called *"contr_X"* as it is the contribution of the source under 
study (point or subject) to the definition of the X-th PC.

We start by defining the list of variables that one might want to display

{{
    "```{literalinclude} " + parent_dir
    + "/roottest/uranie/doc/dataserver/use_cases/" + language + "/dataserverPCAExample." 
    + extension 
    + "\n"
    + ":language: " + language + "\n"
    + ":lines: " + bloc2[language] + "\n"
    + "```"
}}

From there, once the variable ntuple is retrieved, one can dump both the quality and contribution 
coefficients for the variable, here the subjects (it leads to the second and third block in the 
output shown in [](#use_cases_macro_dataserver_pca_example_console)).

{{
    "```{literalinclude} " + parent_dir
    + "/roottest/uranie/doc/dataserver/use_cases/" + language + "/dataserverPCAExample." 
    + extension 
    + "\n"
    + ":language: " + language + "\n"
    + ":lines: " + bloc3[language] + "\n"
    + "```"
}}

Finally, one can do the same for the data points, with the same `Scan` method (which lead to the 
fourth and fifth block in the output shown in [](#use_cases_macro_dataserver_pca_example_console)).

{{
    "```{literalinclude} " + parent_dir
    + "/roottest/uranie/doc/dataserver/use_cases/" + language + "/dataserverPCAExample." 
    + extension 
    + "\n"
    + ":language: " + language + "\n"
    + ":lines: " + bloc4[language] + "\n"
    + "```"
}}

(use_cases_macro_dataserver_pca_example_console)=
## Console

{{
    "```{literalinclude} " + parent_dir
    + "/roottest/build/uranie/doc/dataserver/use_cases/" + language + "/dataserverPCAExample.log"
    + "\n"
    + ":language: none\n"
    + ":name: console_dataserverPCAExample\n"
    + ":lines: " + bloc5[language] + "\n"
    + "```"
}}