Each model has ten independent variables (called A through J, including I). The value of
each variable, xi , satisfies −1<= xi <=1, for i = 1,2,…,10 . The dependent variable Y is
given by Y = f (a,b,…, j) + c . That is, f (a,b,…, j)is potentially a function of up to ten
variables. There is a belief that the model will involve five or fewer independent
variables (but of course, that belief may not be correct). Your assignment is to
determine f (a,b,…, j) by designing an experiment, requesting the values of the
dependent variable generated by the model for a specified set of runs, and analyzing the
data. The experiment that you design consists of a collection of runs. A run is one setting
of (a,b,…, j)
You will create a spreadsheet that contains your Stony Brook identification number, and
a series of lines specifying the runs to be generated. You should send the spreadsheet
(call it request1ID123456789) to the TA at the e-mail account
For example, the spreadsheet for student 1 (who has SB
ID number 123456789) has the format below:
A spreadsheet with the values of the independent variable and a realization of the
dependent variable for those settings will be emailed to you within two business days
after you submit your spreadsheet. For your TA’s convenience, please save your
spreadsheet in a .csv file. You can achieve this by clicking save as .csv file in Excel.
You will be charged for data using the following cost structure:
1. If the values of the independent variables in a run are either -1 or 1, the cost of the
run is 1 point.
2. If there is at least one 0 value in the run and all other independent variables have a
value -1 or 1, the cost of the 0 runs is 3 points.
3. If at least one of the values of an independent variable xi are such that
−1<= xi <=1, xi \= 0, the cost of the run is 5 points for each value different from -1,
0, or 1.
4. The total cost of your first request is the sum of the costs for the runs requested.
5. The cost of a second request for runs is 100 points plus the costs for the runs
requested. The cost of a third request is 400 points plus the sum of the costs for
the runs requested. The cost of a fourth request is 600 points plus the sum of the
costs for the runs requested. You may not make more than four requests for data.
6. There is no limit to the number of runs you can specify in each request. Of course,
you must pay for each run, and there may not be enough time to generate and
analyze more than 3 sets of data.
Your report should include
1. An Introduction Section in which you specify the rationale of your design.
2. A Method section in which you precisely specify the design and conditions for
getting supplemental data if any and your protocol for validating your model.
3. A Results section that includes: 1. your estimated model; 2. a table that specifies
whether each variable does or does not affect the setting of the dependent variable;
3. your analysis of variance table; and 4. your documentation of the validity of the
model that you report. Include all the data that you requested in your final
analysis. Not using all the data requested will be significantly penalized.
4. Your Conclusions and Discussion section. This section includes the limitations of
More specifically, you must indicate whether each of the fourteen variables is in or out of
your estimated model.