From this report, we will look at the decision tree and the expert system which are some of the artificial intelligence techniques. The company will able to understand the structure and how do the decision tree and the expert system work. This will also allow us to know how they may be used in practice in Business. In addition, we will look at some example, which will allow us to know more about the use of these AI techniques.
3.0 Decision Trees
Decision trees are one of the AI techniques that the company can use to improve their profitability. Using the decision trees will help the company to choose between several courses of action. Possible consequences, which include chance event out comes, resource costs and utility can be show in the tree-like graph, the company can use the decision trees to help identify a goal that a strategy most likely to reach in the operation research, specifically in decision analysis. The structure of the decision trees is very effective, it allow us to explore options, and investigate the possible outcomes of choosing those options. In addition, when the company have limited resources, the decision trees will be particularly useful to choose between different strategies, projects or investment opportunities, which the decision trees can help the company to look at the risks and rewards associated with each possible course of action from a balanced picture. Furthermore, the decision trees are easy to understand and modify and by using the decision tree classification algorithm, it can used to compute a decision trees.
3.1 How does it work?
When the company has a decision need to make, they can start a decision tree by drawing a small square on the left to represent the company. To expand the thoughts, draw lines outward from the box and write a short description for each possible solution.
Picture 1) Showing how to draw a basic decision tree
The results will be at the end of each line, the company can draw a small circle if the result raking that decision is uncertain. And draw a square if the result is another decision that the company need to make. And draw a triangle at the end of the line when the company completed the solution.
Picture 2) Shows the meaning of each nodes
When the company keeps on drawn out as many of the possible outcomes and decision, it will lead us to see the original decisions.
3.2 Example
This is an example showing how to calculate the value of each node to find out which possible decision has the highest expected monetary value.
Option
Possible tender prices
Probability of getting contract
MS1
130,000
0.02
(cost 50)
115,000
0.85
MS2
70,000
0.15
(cost 14)
65,000
0.8
60,000
0.95
MS1 & MS2
190,000
0.05
(cost 55)
140,000
0.65"Your company is considering whether it should tender for two contracts (MS1 and MS2) on offer from a government department for the supply of certain components. The company has three options:
If tenders are to be submitted the company will incur additional costs. These costs will have to be entirely recouped from the contract price. The risk, of course, is that if a tender is unsuccessful the company will have made a loss.
The cost of tendering for contract MS1 only is £50,000. The component supply cost if the tender is successful would be £18,000.
The cost of tendering for contract MS2 only is £14,000. The component supply cost if the tender is successful would be £12,000.
The cost of tendering for both contracts MS1 and contract MS2 is £55,000. The component supply cost if the tender is successful would be £24,000."
Source: Decision tree example 1995 UG exam, OR-Notes, J E Beasley
From the sample data above, we can see that this will be how the decision looks. (All figures in £'000).
Table 1) decision trees of the example.
Source: Decision tree example 1995 UG exam, OR-Notes, J E Beasley
From the decision trees above, we can calculate the value of each possibility decisions. First we need to calculate the total profit of each node.
Total Profit = Possible Tender - Cost - Component Supply Cost
Total profit of MS1:
MS1 cost: 50, Component supply cost: 18
path to terminal node 12
Total profit 130-50-18 = 62
path to terminal node 13,
Total profit -50
path to terminal node 14
Total profit 115-50-18 = 47
path to terminal node 15
Total profit -50
Total profit of MS2:
MS2 cost: 14, Component supply cost: 12
path to terminal node 16
Total profit 70-14-12 = 44
path to terminal node 17
Total profit -14
path to terminal node 18
Total profit 65-14-12 = 39
path to terminal node 19
Total profit -14
path to terminal node 20
Total profit 60-14-12 = 34
path to terminal node 21
Total profit -14
Total profit of MS1 & MS2:
MS1 & MS2 cost: 55, Component supply cost: 24
path to terminal node 22
Total profit 190-55- 24=111
path to terminal node 23
Total profit -55
path to terminal node 24
Total profit 140-55- 24=61
path to terminal node 25
Total profit -55
The table below can show the total profit of each terminal node.
Terminal node
Total profit
12
62
13
-50
14
47
15
-50
16
44
17
-14
18
39
19
-14
20
34
21
-14
22
111
23
-55
24
61
25
-55
Table 2) Total Profit of each terminal node
Then we can work out which decisions will have the highest expected monetary value by working out the EMV. Then choose the nearest EMV value to the possibility tender prices (lowest).
EMV = (Probability of Getting Contract) X (Highest Total Profit)
+
(100% - Probability of Getting Contract) X (Lowest Total Profit)
MS1:
Possible tender Price: 115
node 5 : EMV is 0.2(62) + 0.8(-50) = -27.6
node 6 : EMV is 0.85(47) + 0.15(-50) = 32.45
MS1 EMV is 32.45
MS2:
Possible tender Price: 60
node 7 : EMV is 0.15(44) + 0.85(-14) = -5.3
node 8 : EMV is 0.80(39) + 0.20(-14) = 28.4
node 9 : EMV is 0.95(34) + 0.05(-14) = 31.6
MS2 EMV is 31.6
MS1 & MS2:
Possible tender Price: 140
node 10 : EMV is 0.05(111) + 0.95(-55) = -46.7
node 11 : EMV is 0.65(61) + 0.35(-55) = 20.4
MS1 & MS2 is 20.4
The table below shows EMV in MS1, MS2 and MS1&MS2
Tender
EMV
MS1
32.45
MS2
31.6
MS1&MS2
20.4
Table 2) EMV of each tender
From the calculation above, we can see that MS1 (at a price of 115) as it has the highest expected monetary value of 32.45. It also shows us that by using the decision tree, we can find the highest expected monetary node. It will help the company to choose and make decision more easily and shows the EMV of each node thought the development. We can see which decision will be the more profitable or which decision will not, so it can help the company to make decision and choose the right decision to make more profit.
Advantages
Simple to understand and interpret
Have value even with little hard data
Can be combined with other decision techniques
4.0 Expert System
One of the largest areas of applications of artificial intelligence is in expert system; it is software that provides an answer to a problem which uses non- numeric knowledge to solve problems. Exploiting the specialized skills or information that held by a group of people on specific areas is what this system will do as it can be though of a computerized consulting service. Expert system can also be called an information guidance system; this system can also be use in the company to provide financial guidelines.
4.1 How does it work?
The structure of an expert system includes:
The knowledge base
Reasoning used to hold the set of rules of inference. IF-THEN rules are used in most of the systems to represent knowledge. Typically systems can have from a few hundred to a few thousand rules.
The database
It is generally considered to be a set of useful facts that gives the context of the problem domains. As the IF THEN rules can be thought of, these are facts satisfy the condition part of the condition action rules.
The rule interpreter
Often known as an inference engine, it uses the set of facts to produce even more facts from controlling the knowledge base. Providing a natural language interface which is the ideally result of communication with the system. This enables a user to interact independently of the expert with the intelligent system.
Source: EXPERT SYSTEMS http://www.cs.cf.ac.uk/Dave/AI1/mycin.html
4.2 Example
The company can use the expert system in their financial management. This is a example of financial expert system in a company.
"action finances ;
do restart
and spy_fact(fund)
and spy_fact(for_savings)
and spy_fact(for_investment)
and nl
and write( 'started') and nl
and ask fund
and invoke ruleset finance_rules
and write( 'finished ' )
and nl .
ruleset finance_rules
contains all rules;
update ruleset by removing each selected rule .
question fund
please enter the amount of your fund;
input integer.
rule low_funds
if
fund < 1000
then
for_savings becomes fund
and for_investment becomes 0
and remember that for_savings
and remember that for_investment.
rule good_funds
if fund > 1000
then
for_savings becomes 1000
and for_investment becomes fund - 1000
and remember that for_savings
and remember that for_investment."
Source: Lecture note week 5, Steven Northam, 2010
From this example, when a company type in a number, the expert system will suggest how much will the company should save and how much should the company used for investment. More over, a typical expert system can be more complicated and ask more questions, which will lead us to a more accurate and reliable answer. The expert system can help the company to due with the financial question by giving suggestions and the company will able to use the money usefully. For example, if the company have a income less then 1000 this month, the expert system will suggest the company use less money for investment this month, and maybe less savings which will according what the company wants.
5.0 Conclusion
Using the decision tree, we can find the highest expected monetary decision node. It will help the company to choose and make decision more easily and shows the EMV of each node thought the development. We can see which decision will be the more profitable or which decision will not, so it can help the company to make decision and choose the right decision to make more profit.
The expert system can help the company to due with the financial question by giving suggestions and the company will able to use the money usefully. The expert system will suggest how much will the company should save and how much should the company used for investment. The more complicated and detail questions/ expert system will lead to the higher accuracy and more reliable answer.
The company can use the decision trees with the expert system; decision trees can help to make the most suitable decisions and questions, which will help to use the expert system more easily.
