# Difficulty of Complicated Models

Many methods of
Multi-Variable Analysis
can be used in the real world today.

There are two types in these methods.

One is the model that deals with the complicated relations among variables.
For example,
SEM and Covariance Structure Analysis
,
Bayesian Network
and
Associations Analysis
.

The other is the model that one Y variable and many X variables.
For example,
Support Vector Machine
and
Neural Network
.

##
Find Arrows

In classical methods, the analyst has to set the arrows.

But in
Bayesian Network
and
Associations Analysis
, the method find arrows automatically.

So these tools are strong as the method of
Statistical Way of Making Hypothesis
.

We need to understand that the arrows are find by
Asymmetry of Quantity Data
.
And the methods do not find the rules of causes and effects.

##
Non Linear Relationship

The basic models deals with
Linearity
.
But
Support Vector Machine
and
Neural Network
can deals with non-linearity.

##
Difficulty of Complicated Models

The difficulty of complicated models have complicated difficulty of classical models.

So it is not easy to use these models.

###
Use of Complicated Models

The output of complicated models cannot be used as the solution directly.

I often use complicated models to get the **hints** to solve problems.
I also use classical models and graphical approaches.

There are two types of complicated models.
And maybe there is the hybrid model of the two.
But we need to avoid various faults of analysis.

NEXT Robust Analysis