Data Analysis 101

1
Data Analysis 101
Bill Atwood SCIPP/UCSC Instrument Analysis
Workshop II 27-Sept-2004
2
Modes of Working
Recon Root Files All the information not a
tabular structure - All the tracks - All the
hits on each track - Etc. Large Files Root
Macros nTuple Key information distilled from
Root Files Flat Tabular structure (Xcel Spread
Sheet) Large number columns to make
possbile an "in-depth" view of data SVAC
Some of both
3
What Each Mode can be used for
Root Files Full analysis - Projection of
tracks into subsystem of interest - In case of
Gleam environment Full power of
geometry (no hard coding of numbers!) nTuple
- Many parameters of track trajectories
present. Allows of simple projections into
subsystem. (geometry must be
entered by hand!) - Allows for correlation
between variables to be studied.
4
My own style - Develop Plots of interest using
nTuple Often, must create new variables,
make projections, etc. -
Most variables of interest should not depend
on location or angles Image the
variable Explore using scatter plots, profile
plots, etc. - Back results up into the Root
oriented Analysis Much of the
Analysis Package evolved in this manner.
5
Analysis Package Purpose Provide subsystem
and system wide parameters for
Event Analysis leit-motif Provide a collection
of templates and examples on
how to use the system (TDS Services Tools).
A set of classes where each class focuses on
calculating variables associated with a
subsystem - or several together All have
names like McValsTool, CalValsTool,
TkrValsTool,...
6
Image Example
Require Tkr1Gaps gt 0 This is was the
original idea Image the Tracker Layout
7
Collapse down onto one Tower Cut
off the edges ( 8 mm)
Collapse down onto one SSD
Cut off the edges ... ( 4 mm)
These are the events that shouldn't be... Lets
Explore!
8
Require 1 Track
Require Small c2
Good Events
InEff. Events
Require a MIP
9
This suggests we require 1) Tkr1Chisq lt
.2 2) .8 lt CalMIPRatio lt 1.2 3)
TkrNumTracks 1 Leaves 4.1 K Events ( 40
Efficiency)
This leaves 1 Ineff. Event
10
Goal No Hard-Coded Geometry
Dual use of geometry Monte Carlo Reconstruction
Method of Accessing Geometry Service -
Give global constants (size of tower, etc.)
Propagator - Projects trajectories through 3D
geometry - Allows for error propagation,
material audits, etc.