These two
Attribution Analysis 81
Return
Time
Return
Liquidity
Time
Liquidity
Return
Risk
Time
Risk
Liquidity
Risk
Return
Time
Liquidity
Return
Time
Risk
Time
Liquidity
Risk
Return
Risk
Liquidity
Return
Time
Liquidity
Risk
Level 1
Level 2
Level 3
Level 4
Return Time Liquidity Risk
1
2
3
1.a
2.a
3.a
Core Attributes
Converging Level
Requirements
EXHIBIT 4.6 FORWARD ATTRIBUTE SELECTION METHOD
categories will provide different values to investors, if indeed such services are to be built. The
first will be centered on the return on investment policy, whereas the second accentuates liquidity
of assets during the investment period.
BACKWARD ATTRIBUTE ELIMINATION TECHNIQUE. This method advocates starting from the
lower levels of the attribution model network and moving up toward level 1 to enable proper
filtering of attributes. At which node should one start searching the network? It is recommended
to start from the very bottom, from the converging node, and move up to higher levels. This
elimination technique enables the operator to filter out any attributes that may not contribute to
the solution domain and are not significant product descriptions.
Exhibit 4.7 exemplifies two backward attribution elimination scenarios in which attributes
are filtered out in each upward step. These backward search paths lead to the final nodes: 3,
and 2.a, ending at level 2 and level 3, respectively.
Pages:
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160