Guide for Mac Excel 2011
Decision Analysis with TreePlan
TreePlan is an Excel add-in that enables you to create decision trees in an Excel spreadsheet.
This guide to using TreePlan consists of the following parts:
1. Getting Started
2. Adding Nodes
3. Copying Subtrees
4. Interpreting the Results
5. Sensitivity Analysis
We use the TreePlan software to build a decision tree for the first part of the Wildcat Oil
Drilling case. This part of the case concerns the decision as to whether or not to drill. The
cost of drilling the site is $70m. If the well is “dry”, there will be no revenue. If it is “wet”,
revenue will be $220m. If “soaking”, revenue will be $670m. If there is underlying lime-
shale rising into a flat dome shape, the chances of finding oil increase. However, we don’t
know yet if there is a dome. There is an estimate of 6 chances in 10 of finding a dome on the
current site. The following table gives conditional probabilities of the drilling result, given
the existence or otherwise of the dome. For example, 85% is the probability of the well being
dry, given that there is no underlying dome structure.
Dome No Dome
Dry Well 0.60 0.85
Wet Well 0.25 0.125
Soaking Well 0.15 0.025
1.00 1.00
1
1. Getting Started
To check that the software is available, when you click Tools, you should see “Decision
Tree” or “TreePlan” at the bottom of the drop-down menu.
To get started with a new decision tree, open a new empty Excel spreadsheet, and position the
cursor in cell A1. Click Tools, and select Decision Tree.
After the dialog box relating to the licensing conditions, in the following dialog box, select
New Tree:
The following simple tree appears:
Click on cell D2, which contains the label “Alternative 1” for the top branch. Type “Drill”.
Click on cell D7, which contains the label “Alternative 2” for the lower branch. Type “No
Drill”.
In TreePlan, costs and payouts can be specified throughout the tree. Indicate the cost of
drilling by typing “-70” in cell D4. You should now have the following tree:
Make sure to save your spreadsheet frequently as you build the tree.
2
2. Adding Nodes
Now we need to include a chance node following the Drill branch of the tree. This chance
node must have two branches representing whether or not there is a dome.
Click on cell F3, which contains the end (triangular) node of the Drill branch. Click Tools,
and select Decision Tree. In the ensuing dialog box, select Change to event node, and
specify the number of Branches as Two, and click OK.
The tree should appear as:
Change the branch labels “Outcome 3” and “Outcome 4”, in cells H2 and H7, to “Dome” and
“No Dome”, respectively.
Change the probability on the “Dome” and “No Dome” branches (from 0.5) to 0.6 and 0.4,
respectively.
Leave the costs on these two branches as 0.
The tree should be of the following form:
3
Next, we need to add a new chance node to the end node of both the Dome and No Dome
branches. This new chance node represents the three possible drilling results, Dry, Wet and
Soaking.
Click on cell J3, which contains the end node of the Dome branch. Click Tools, and select
Decision Tree. In the ensuing dialog box, select Change to event node, and specify the
number of Branches as Three, and click OK.
For the chance node that follows the Dome branch:
(i) change the branch labels to “Dry”, “Wet” and “Soaking”;
(ii) change the probabilities on the “Dry”, “Wet” and “Soaking” branches (from 0.333333) to
0.6, 0.25 and 0.15, respectively; and
(iii) change the values on the “Dry”, “Wet” and “Soaking” branches (from 0) to 0, 220 and
670, respectively.
The new tree is of the form:
4
3. Copying Subtrees
To the end node of the No Drill branch of the tree, we now need to add a similar Drilling
Results chance node to the one that we have just created. A convenient way to do this is to
copy the chance node that we have just created, and then edit appropriately the probabilities
in the new node.
To copy the chance node that we have just created, click
on cell J8, which contains that node. Click Tools, and
select Decision Tree. In the ensuing dialog box, select
Copy subtree, and click OK.
Now click on cell J18, which contains the end node of the
‘No Dome’ branch of the tree. Click Tools, and select
Decision Tree. In the ensuing dialog box, select Paste
subtree, and click OK.
For the chance node that follows the No Dome branch, change the probabilities on the “Dry”,
“Wet” and “Soaking” branches to 0.85, 0.125 and 0.025, respectively. Do not change the
labels or values on these branches.
The completed tree should be of the form:
5
4. Interpreting the Results
TreePlan automatically calculates the cumulative value for each path through the tree. These
are displayed at the extreme right end of the tree.
Expected values are presented just below each chance and decision node.
The optimal course of action at the decision node, according to an expected value criterion, is
indicated with a number inside this node. For the oil drilling decision tree in this guide, the
number in the node is 1, indicating that the optimal decision corresponds to the first branch
coming from the decision node. The optimal course of action is to drill, and this has an
expected value of 41.
For the oil drilling decision tree in this guide, the optimal decision is the one that maximises
the expected value (profit). For some other trees, it can be appropriate to minimise expected
value (costs). The criterion used to evaluate the tree can be changed in TreePlan by first
clicking on an empty cell in the spreadsheet containing the decision tree. Then click Tools,
and select Decision Tree. In the ensuing dialog box, select Options. The next dialog box
allows you to select Maximize (profits) or Minimize (costs) for the expected value (EV).
6
5. Sensitivity Analysis
5.1. One-Way Sensitivity Analysis based on Excel’s Data Tables
It can be useful to perform sensitivity analysis for various parameters in the tree. As an
example, let’s perform a one-way sensitivity analysis on the Soaking Revenue, which we
have so far assumed to be 670. Well, this value appears twice in the decision tree. To ensure
that our sensitivity analysis is performed on both cells that contain the Soaking Revenue, we
need link the two cells. In my worksheet, this involves changing the entry in cell L29 from
‘670’ to ‘=L14’, as shown below.
We now use Excel’s Data Table feature to perform a sensitivity analysis on the Soaking
Revenue value, which is in cell L14 of my worksheet. Note that Data Table is not part of
TreePlan, and is always available in Excel.
In the current tree, we have soaking revenue of 670 leading to an expected value of 41 at the
decision node. Using an Excel Data Table, we will now evaluate the expected value for a
variety of different values of soaking revenue.
In a column, in an empty part of the worksheet, list values of soaking
revenue for which you require expected values. At the top of the next
column, you need to refer to the expected value cell, which is A25. This
initial table should be of the following form:
Next we complete the table in one operation using the Data Table feature.
7
Highlight the table of values as shown:
From the main Excel menu, select Data, and from the drop-down menu select Data Table.
In the Data Table dialog box, specify Column input cell as L14. This
is the cell in the decision tree where soaking revenue is located. Click
OK. The table should now be filled with the expected value for each
of the soaking revenue values.
Using an Excel scatterplot, we can plot these values to show the expected value rising with an
increase in the soaking revenue.
8
5.2. Two-Way Sensitivity Analysis based on Excel’s Data Tables
Now let’s perform a two-way sensitivity analysis for the Soaking Revenue and the Drilling
Cost.
In the current tree, we have soaking revenue of 670 and drilling cost of 70 leading to an
expected value of 41 at the decision node. Using an Excel Data Table, we will now evaluate
the expected value for a variety of different values of soaking revenue and drilling cost.
Set up an initial table with potential values of soaking revenue down the left hand side and
potential values of drilling cost across the top row. In the top left cell, you need to refer to the
expected value cell, which is A25. This initial table should be of the following form:
Next we complete the table in one operation using the Data Table feature.
Highlight the table of values as shown:
From the main Excel menu, select Data, and from the drop-down menu select Data Table.
In the Data Table dialog box:
(i) Specify Row input cell as D16. This is the cell in the decision tree
where drilling cost is located.
(ii) Specify Column input cell as L14. This is the cell in the decision
tree where soaking revenue is located.
(iii) Click OK.
9
The table should now be filled with the expected value corresponding to each pair of soaking
revenue and drilling cost values.
To chart this table of values, I would recommend copying the table, and then using paste
special to paste values into an empty space in the worksheet. Now delete the value of 41 in
the top left cell of the table. This new table can be plotted using a line chart as:
10
