Identifying Data and Methods Assignment | Get Paper Help
Introduction
To proceed with your analysis, you’ll determine which method you will use to analyze your data and
quantify the parameters and other functions in your model. Note, if you have already identified the
analytic method you will use, you do not require this guide.
Step 1. Identify Whether You Will Be Able to Complete Your Analysis
To identify whether you can move ahead with your complete analysis within the course of this program,
consider whether you face one of the following three situations. Keep in mind that you will have one
week to complete your analysis:
1. The analysis required is simple: For example, after framing your analysis, you might determine
that you only need to quantify some critical ratios your organization has not previously
considered. This might be the case when you want to identify relationships under different
conditions (e.g., when you gather data through an A/B test). Remember how Professor Dubé
was able to support ZipRecruiter using some basic graphs after they ran their experiment?
When the parameters you must quantify are simple, you can proceed without statistical
support.
2. The analysis may be moderately complex, such as a linear regression, but you have
quantitative background: For example, your model may be framed as a linear equation. To
quantify the parameters in a linear model, the typical method is to run a linear regression (For
more background on linear regression, see here). You may have the skillset required to do this
analysis or support from others who do and so, likewise, can proceed.
3. The analysis may be more complex, but you have support from others: In the extreme, you
may require complex methods like the ones the faculty discuss in some of their examples. For
example, Professor Misra’s example about sales compensation required complex non-linear
relationships to develop the “Virtual Salesperson.” You may be able to proceed directly ahead
even if this case if you have ready access to data and the support of data professionals.
Remember, the goal of this program is to help you lead data & analytics efforts, not to become a data
scientist. If you require additional support to gather the data you’ll need or identify or execute the
appropriate method, then do not proceed to complete your analysis. Instead, use the following section
to identify how to test your model during the program. At the end of the program, you will then propose
whether to mobilize the resources needed to refine your results via a more intensive analysis.
Step 2: Capture Your Conclusion in your Workbook
If you can complete your analysis, indicate so in your workbook along with the method you will use
(e.g., run simple statistics, conduct a linear regression, …).
If you cannot, indicate in your workbook that you will test your model and move ahead to the next
section.
Guide to Identifying Methods
ExecOnline – Leading Data & Analytics
Step 3. Identify How You Will Test Your Model
Use this section of the guide if you have concluded that you will test your model during the next
module.
Business leaders often take exploratory first steps before making an investment to identify whether it
will generate a return. If you’ve reached here, you’ve identified that you’ll need to invest time or
resources after the program to complete your analysis. To validate that doing so will generate a return,
identify how you will test your model during the next module.
Below, we offer three common ways business leaders can test out their models. Doing so can not only
provide evidence to support investing in a complete data analysis, but also provide initial insights that
may point the way toward business decisions. Choose one or more that you will plan to pursue in the
next module and capture your decision in your workbook.
Test Method 1: Plot Your Data
Your model may feature non-linear correlations between factors. While those may be difficult to
quantify statistically, they often can show up quite clearly visually. Remember the simple graphs that
Professor Dubé used to illustrate the ZipRecruiter results? They faced non-linear demand versus price.
But the graphs made it easy for leadership to see the impact of different choices.
In this method, you plot the data that represent the correlations you wish to quantify. Then, you can
manually draw in a line or curve that summarize the data. While this graph will be approximate, you can
use it to estimate a value for a parameter, identify additional data points to run through your model, or
locate critical points where returns begin to diminish or relationships between factors change.
Test Method 2: Identify Upper & Lower Bounds
You may require complex analysis to exactly quantify your parameters. At the same time, you may be
able to identify bounds for them with confidence. Can you do this for your parameters? What’s the
largest they might be based on your experience? The smallest?
Note: This approach can work in tandem with plotting your data. You may wish to first plot the data you
have and then draw two curves representing the upper and lower bounds you’ve identified.
Test Method 3: Work Backwards to Identify Boundaries for Success
Even if you struggle to quantify your model through the approaches above, you may be able to work
backwards. What levels must your parameters have to motivate you to recommend one path forward
over another? For example, if you are determining whether to invest in search advertising, what is the
minimum return you’d expect on that investment? What is the minimum percentage of prospects that
would need to click on ads to hit that metric? Your model gives you the structure to run such back-ofthe-
envelope thought experiments.
With such “boundaries for success” in hand, you can examine the data you have. How likely is it that the
actual parameters fall above or below where they would need to be?
Introduction
To proceed with your analysis, you’ll determine which method you will use to analyze your data and
quantify the parameters and other functions in your model. Note, if you have already identified the
analytic method you will use, you do not require this guide.
Step 1. Identify Whether You Will Be Able to Complete Your Analysis
To identify whether you can move ahead with your complete analysis within the course of this program,
consider whether you face one of the following three situations. Keep in mind that you will have one
week to complete your analysis:
1. The analysis required is simple: For example, after framing your analysis, you might determine
that you only need to quantify some critical ratios your organization has not previously
considered. This might be the case when you want to identify relationships under different
conditions (e.g., when you gather data through an A/B test). Remember how Professor Dubé
was able to support ZipRecruiter using some basic graphs after they ran their experiment?
When the parameters you must quantify are simple, you can proceed without statistical
support.
2. The analysis may be moderately complex, such as a linear regression, but you have
quantitative background: For example, your model may be framed as a linear equation. To
quantify the parameters in a linear model, the typical method is to run a linear regression (For
more background on linear regression, see here). You may have the skillset required to do this
analysis or support from others who do and so, likewise, can proceed.
3. The analysis may be more complex, but you have support from others: In the extreme, you
may require complex methods like the ones the faculty discuss in some of their examples. For
example, Professor Misra’s example about sales compensation required complex non-linear
relationships to develop the “Virtual Salesperson.” You may be able to proceed directly ahead
even if this case if you have ready access to data and the support of data professionals.
Remember, the goal of this program is to help you lead data & analytics efforts, not to become a data
scientist. If you require additional support to gather the data you’ll need or identify or execute the
appropriate method, then do not proceed to complete your analysis. Instead, use the following section
to identify how to test your model during the program. At the end of the program, you will then propose
whether to mobilize the resources needed to refine your results via a more intensive analysis.
Step 2: Capture Your Conclusion in your Workbook
If you can complete your analysis, indicate so in your workbook along with the method you will use
(e.g., run simple statistics, conduct a linear regression, …).
If you cannot, indicate in your workbook that you will test your model and move ahead to the next
section.
Guide to Identifying Methods
ExecOnline – Leading Data & Analytics
Step 3. Identify How You Will Test Your Model
Use this section of the guide if you have concluded that you will test your model during the next
module.
Business leaders often take exploratory first steps before making an investment to identify whether it
will generate a return. If you’ve reached here, you’ve identified that you’ll need to invest time or
resources after the program to complete your analysis. To validate that doing so will generate a return,
identify how you will test your model during the next module.
Below, we offer three common ways business leaders can test out their models. Doing so can not only
provide evidence to support investing in a complete data analysis, but also provide initial insights that
may point the way toward business decisions. Choose one or more that you will plan to pursue in the
next module and capture your decision in your workbook.
Test Method 1: Plot Your Data
Your model may feature non-linear correlations between factors. While those may be difficult to
quantify statistically, they often can show up quite clearly visually. Remember the simple graphs that
Professor Dubé used to illustrate the ZipRecruiter results? They faced non-linear demand versus price.
But the graphs made it easy for leadership to see the impact of different choices.
In this method, you plot the data that represent the correlations you wish to quantify. Then, you can
manually draw in a line or curve that summarize the data. While this graph will be approximate, you can
use it to estimate a value for a parameter, identify additional data points to run through your model, or
locate critical points where returns begin to diminish or relationships between factors change.
Test Method 2: Identify Upper & Lower Bounds
You may require complex analysis to exactly quantify your parameters. At the same time, you may be
able to identify bounds for them with confidence. Can you do this for your parameters? What’s the
largest they might be based on your experience? The smallest?
Note: This approach can work in tandem with plotting your data. You may wish to first plot the data you
have and then draw two curves representing the upper and lower bounds you’ve identified.
Test Method 3: Work Backwards to Identify Boundaries for Success
Even if you struggle to quantify your model through the approaches above, you may be able to work
backwards. What levels must your parameters have to motivate you to recommend one path forward
over another? For example, if you are determining whether to invest in search advertising, what is the
minimum return you’d expect on that investment? What is the minimum percentage of prospects that
would need to click on ads to hit that metric? Your model gives you the structure to run such back-ofthe-
envelope thought experiments.
With such “boundaries for success” in hand, you can examine the data you have. How likely is it that the
actual parameters fall above or below where they would need to be?
