Abstract
The treelet transform (TT) is a recent data reduction technique
from the field of machine learning. Sharing many similarities with principal
components analysis (PCA), TT can reduce a multidimensional data set to
the projections on a small number of directions or components which account
for much of the variation in the original data. However, in contrast to PCA,
TT produces sparse components. This can greatly simplify interpretation. We
describe the tt Stata add-on for performing TT. The add-on includes a Mata
implementation of the TT algorithm, alongside other functionality to aid the
practical application of TT. We show how a basic exploratory data analysis
using the tt add-on might look.
from the field of machine learning. Sharing many similarities with principal
components analysis (PCA), TT can reduce a multidimensional data set to
the projections on a small number of directions or components which account
for much of the variation in the original data. However, in contrast to PCA,
TT produces sparse components. This can greatly simplify interpretation. We
describe the tt Stata add-on for performing TT. The add-on includes a Mata
implementation of the TT algorithm, alongside other functionality to aid the
practical application of TT. We show how a basic exploratory data analysis
using the tt add-on might look.
Original language | English |
---|
Publisher | Department of Mathematical Sciences, Aalborg University |
---|---|
Number of pages | 14 |
Publication status | Published - 2011 |
Series | Research Report Series |
---|---|
Number | R-2011-09 |
ISSN | 1399-2503 |
Keywords
- treelet
- PCA
- dimension reduction
- factor analysis