Case Study #4

Dataset name: User Identification From Walking Activity

Data Dataset location: http://archive.ics.uci.edu/ml/machine-learning-databases/00286/User%20Identification%20From%20Walking%20Activity.zip

Dataset attributes: 149332 observables of 4 variables; accelerometer data from 22 subjects

Dataset info: https://archive.ics.uci.edu/ml/datasets/User+Identification+From+Walking+Activity Problem: Classify and identify subjects based on motions

In this assignment, you will demonstrate your mastery of the following course outcomes:

Analyze machine learning algorithms against sample data for determining algorithm operations and beneficial impacts [IT-460-01]

Evaluate tools that implement machine learning algorithms for their risks and benefits to information technology organizations [IT-460-02]

Analyze machine learning models that utilize sample data by employing the standard practitioner cycle [IT-460-03]

Determine the impact that machine learning techniques can have on organizational tasks and problems by analyzing specific model uses [IT-460-04]

Develop and present ethical use policies about information derived from organizational machine learning activities [IT-460-05]

Prompt

Your model evaluation and ethical use policy should answer the following prompt: Select a case and machine learning algorithm from those provided and analyze the use of that model and the accompanying data for the problem at hand, comment on the accuracy and value of that model for the organization, and craft a general ethical use policy for the model. Specifically, the following critical elements must be addressed: I.

Model and Dataset Identification and Defense a) Introduction: Introduce the algorithm, case, and data that you have selected for your project by explaining the problem or task at hand and the expected solution or outcome of the use of the model. In other words, when your implementation of the model is done, what is the desired result for the organization? [IT-460-01]

b) Algorithm/Model Evaluation: Why is the algorithm/model used for the case you have selected the most appropriate model for the data? Do you feel that another model would be more appropriate to run the data? If so, which one and why? Whether you feel the model is the most appropriate or not, be sure to thoroughly explain why, providing specific examples or external references to help clarify your stance. [IT-460-01]

c) Value to the Organization: How will the use of this model add value to the organization? In other words, what is the importance and potential impact of the use of this model for the organization as a whole? Be sure to cite specific examples for support. [IT-460-01]

d) Use of Tool: How will you use the tool mandated to execute the model? Provide the step-by-step process you will use. [IT-460-02]

e) Tool Evaluation: Evaluate the tool being used for its benefits and shortcomings in considering the specific information from the scenario. Are there alternatives that you would suggest? Are there shortcomings or clear benefits as compared to other tools available in the industry? Your evaluation of the mandated tool and alternative tools should be supported with examples and resources where appropriate. [IT-460-02] II.

Model Execution a)

Initial Execution: Run/train the model with the given initial parameters on the test dataset and explain your process and results. (Note: You should include your results in your report, formatted in the manner most appropriate for clarity.) [IT-460-03]

b) Initial Evaluation: Evaluate these initial conditions compared to the validation dataset from the scenario and your expectations of the results. How do your results meet up with the validation dataset? [IT-460-03]

c) Parameter Changes: Change parameters to investigate the effects to the model accuracy. How did changing the parameters impact your results? Are there trends or patterns that you notice? [IT-460-04]

d) Parameter Confirmation: Identify the parameters you feel are best for this model and dataset and explain what led you to this conclusion. Be sure to explain your reasons clearly, calling on specific examples to assist your explanation. [IT-460-04]

e) Organizational Impact: Given your parameter selection and model execution, what are the impacts on the organizational task or problem at hand? How could the model and results be used to impact the organization? [IT-460-04]

III. Ethical Use Policy

a) How-To: Develop guidelines for the organization that outlines recommended techniques and steps for ensuring accuracy of the model in reuse/alternative uses. Consider the entire practitioner cycle in your guidelines, including the steps that were already completed for you upfront. In other words, while you did not perform the beginning steps of the practitioner cycle, because the data was provided for you, make sure you specify these steps in your guidelines for your organization. [IT-460-03]

b) Ethical Considerations: Analyze the scenario you chose to attend to in your project and determine the potential ethical issues that could arise during any part of the practitioner cycle. What would you do to ensure that those following the guidelines that you developed stayed within ethical bounds? When and how could practitioners potentially go astray in following ethical standards throughout creation and use of a model? There are no right or wrong answers here, so you will not be graded on whether or not you are correct. Instead, you will be assessed on whether you have identified possible ethical concerns and whether you are able to suggest ways to keep practitioners in line with what you feel is ethical practice. [IT-460-05]

c) Ethical Use: What are the potential risks of unethical use of your results within your organization? What would you suggest to ensure that the results from your use of the model were used in an ethical manner? Explain your reasoning and concerns using examples. [IT-460-05]