Vitamin Information Center

Canonical Correlation Analysis
Definition
A canonical correlation analysis is a multivariate statistical
technique that investigates the relationships
between two (or more) sets of variables. In most applications,
however, the two sets are not treated symmetrically;
rather, one set is the predictor set, which is
the set of independent variables, and the other set is
the response set, which is the set of dependent variables.
For example, one may want to study the relationship
of various risk factors to the development of
a group of symptoms; compute the (simultaneous) relationship
between three measures of scholastic ability
with five measures of success in school; or investigate
the relationship between two predictors of social mobility
based on interviews, with actual subsequent social
mobility measured by four different indicators. The
underlying principle is to develop two linear combinations
(i. e., canonical variables) of variables in each set
(both dependent and independent if such a distinction is
made) that best explain the variation in the variables of
the other set, i. e. such that the correlation between the
composite variates is maximized.

Artificial Neural Network
Synonyms
Neural network
Definition
An analytic modeling technique modeled after the
(hypothesized) processes of learning in the cognitive
system and the neurological functions of the brain.
It is capable of predicting new observations (on specific
variables) from other observations (on the same
or other variables) after executing a process of socalled
learning from existing data. Artificial neural networks
(ANN) are nonlinear and capable of modeling
extremely complex functions by creating connections
between processing elements – the computer equivalent
of neurons. For example, the onset of a particular
medical condition could be associated with a very
complex (e. g., nonlinear and interactive) combination
of changes on a subset of the variables being monitored
(e. g., a combination of heart rate, levels of various
substances in the blood, respiration rate). Neural
networks have been used to recognize this predictive
pattern so that the appropriate treatment can be prescribed.
A distinction can be made between two different
types of ANN– networks designed for supervised
learning tasks (e. g., Multilayer Perceptron, Bayesian
networks, Genetic algorithms) and networks primarily
designed for unsupervised learning (Self Organizing
Feature Map (SOFM, or Kohonen) networks).

Artificial Neural Network
Synonyms
Neural network
Definition
An analytic modeling technique modeled after the
(hypothesized) processes of learning in the cognitive
system and the neurological functions of the brain.
It is capable of predicting new observations (on specific
variables) from other observations (on the same
or other variables) after executing a process of socalled
learning from existing data. Artificial neural networks
(ANN) are nonlinear and capable of modeling
extremely complex functions by creating connections
between processing elements – the computer equivalent
of neurons. For example, the onset of a particular
medical condition could be associated with a very
complex (e. g., nonlinear and interactive) combination
of changes on a subset of the variables being monitored
(e. g., a combination of heart rate, levels of various
substances in the blood, respiration rate). Neural
networks have been used to recognize this predictive
pattern so that the appropriate treatment can be prescribed.
A distinction can be made between two different
types of ANN– networks designed for supervised
learning tasks (e. g., Multilayer Perceptron, Bayesian
networks, Genetic algorithms) and networks primarily
designed for unsupervised learning (Self Organizing
Feature Map (SOFM, or Kohonen) networks).

We’re still in a recession. Our currency is still struggling. Inflation is worsening. Companies are worried. I addressed all this last month. This time I want to extend the discussion about where to put technology dollars and how to pitch a recessionary budget. Lots of technology gurus and pundits talk about the distinction between vitamin pills and pain killers. Put another way, there’s an obvious distinction between investments that make money and those that save money – assuming, of course, that all goes well with the projects designed to save or make money. At the simplest level is the relationship between cost management investments (pain killers) versus investments designed to generate revenue (vitamin pills).

So what will it be today – and for the next year or two? (Yes, I think that this recession will be longer and deeper than the last few.) Pain killers – without a doubt.

Those projects that contribute to cost reduction will be well received in many companies, while those designed to generate revenue – vitamin pills – will be closely scrutinized. Continuing with some other metaphors, it may be better to buy or sell shovels as everyone looks for gold-plated growth.

So is it time to buy shares in Cisco and EMC or Oracle and SAP? If you equate infrastructure with pain killers then Cisco and EMC make sense; if you see Oracle and SAP as vitamin pills then you may want to think twice about your investments in their products and services. Conclusion? Infrastructure projects are closer to pain killers and pain killers are closer to cost management than applications projects, vitamin pills or revenue generating projects.

The only problem with all this is that it’s all wrong.

While infrastructure projects often save money (and reduce pain), applications projects are where struggling companies should place their bets. Why? For the same reason why struggling companies in general should play offense rather than defense. This means that while cost management projects are always appealing, when times are tough companies should invest in projects that will eventually make them money. Throwing a long pass on fourth and one is risky – but courageous and often very smart.

As I said last month, there’s no question that projects must either save money or make money in a recessionary economy. But the relative distribution of these projects should be closely scrutinized. The conservative play is to invest only in projects that save you money, often in infrastructure projects that reduce some form of pain. But the smart play is to mix investments with as many – or more – applications projects that can contribute to revenue growth, so-called vitamin pills.

It’s sort of like contrarian stock market investors or those that play counterintuitive trends. Does it work? More often than any of us care to admit. So if you’re about to present your 2008 – 2009 technology budget, here’s some advice:

1. Lead with pain killers likely to yield impressive cost savings; this will build credibility.

2. Selectively propose projects that look like they will save money and make money, knowing that they’re more likely to make money than save it.

3. Mange the hell out of these projects. If they start to go south, kill them quickly and reload with another revenue generating project (in cost-saving clothing).

Some of this requires a little slight of hand and some adult-sized hype. But if you want to emerge from this recession as a hero, then work the system toward projects designed to score revenue gains, not just hold the competition to field goals.