MIS 660 Full Course Discussions GCU

Suppose you wanted to estimate the average household income of all Grand Canyon University (GCU) students. To expedite the process, you only gather household income data from all your friends who major in business at GCU. You then calculate the average income among your friends and report that it represents the average income of all GCU students. Is this a good approach? If not, how would you gather data to derive a better estimate? Explain your answer.

Income data typically have some outliers. For example, Tim Cook, CEO of Apple, Inc., had a salary of about 400 million in 2011. Suppose you had a data set of incomes in 2011 for all GCU faculty and Tim Cook. Which measure of central tendency would you use when reporting on the incomes in your data set if you do not want outliers to have much effect? Explain your answer.

Suppose you had daily temperature data indicating the “high” point of each day for 2015. If you want to show how the high differs over time, what are some of the plot types that will allow you do this? What are some benefits to binning the data into one of 52 weeks and plotting the average high for each week? Would it make sense to do something similar for the four quarters in the year? Why or why not?

Many times, data are missing because of various reasons. This poses some challenges when doing data analysis. For example, suppose you wanted to do some analysis of the yearly incomes of the faculty at GCU. When asked for their incomes, 25% of the faculty did not participate in the survey; therefore, their incomes are missing from the dataset. How would you summarize the income data in this case? Is it appropriate to ignore the missing incomes and summarize the data without them? Should you estimate the missing incomes, perhaps with the overall average, to complete the data set?

Data summarization is usually not enough when performing analysis. Most of the time, adding context by telling a story about the data is necessary to describe the analysis to others, especially those who are not data-savvy. What are some general guidelines to follow to tell a good data story? What story elements or structure should be used to organize the presentation?

Consider your organization, or an organization you are most familiar with. Explain the general process of data aggregation for a typical metric (e.g., sales revenue, cost per unit, etc.) used in the organization. What specific charts are commonly used to visually depict the data? What might be some areas for improvement regarding how the data is visually presented?

What are some of the limitations of using Excel for pivot tables/charts? Why does that make software like Tableau more appealing in the workplace?

Plotting summarized data will almost always help to convey results more easily. However, there are situations where plotting the summarized data instead of creating a simple table makes data interpretation more difficult. Provide two examples of poor charts/graphs and explain why they are difficult to interpret.

Data is useless without the skills to manipulate, summarize, and analyze it. In fact, even after data is summarized into a reporting format such as graphs and tables, it still requires someone to add context and describe the results to fully explain the data. This can be difficult, especially if data is being presented to nontechnical individuals. Describe two techniques that can be used to better describe analysis results to nontechnical individuals.

When most people think about data reporting or visualization, they think about a nicely crafted graph that will not be interactive with a user. Some new tools, such as Tableau, can create visualizations that can interact with a user with informative pop-up information, more drill-down information, and the ability to export filtered results. Describe two benefits to having a user interact with a standard report. Are there any drawbacks if the user modifies the report?

Summarize key data distribution concepts including probability mass functions (PMF), probability density functions (PDF), and cumulative distribution functions (CDF). Based on your organization or any organization you are most familiar with, provide an example of a PMF, an example of a PDF, and an example of a CDF, based on the type of data used in the organization.  How would you summarize each of these to someone who is not familiar with each of these functions?

Suppose you had a six-sided die where each number (1, 2, 3, 4, 5, and 6) has the same probability of showing up (1/6). If the die is rolled an infinite number of times and the number recorded, what will be the average value that shows up? Is the average value one of the actual possibilities (1, 2, 3, 4, 5, or 6)? Why or why not?

Suppose you wanted to understand the relationship between a customer’s yearly income (X) and the number of movies (Y) the customer watched in a year. You then gather data on incomes and the number of movies watched in a year. The range of incomes in your data set is $5K to $150K. After fitting a simple linear model and performing all the appropriate diagnostics, the model showed that, on average, for every $10K in income, the customer watched 1.5 movies in the year. So, for example, if a customer earned 60K in a year, he or she would be expected to watch nine movies during the year. Now you want to apply this model to your very wealthy friend who will earn $1 million in the next year. Is this an appropriate application of your model? Why or why not? Provide specific examples to justify your opinion.

If you regress daily high temperature (Y) on the amount of ice cream sales (X), you will notice that there is a strong positive correlation between the two. In other words, as daily ice cream sales increase, the daily high temperature increases. This implies that if we knew the amount of ice cream sales in a particular day, we could estimate, with a high level of accuracy, the high temperature in that day. Does this mean that if we wanted to increase the daily temperature, we need to sell more ice cream? Explain why or why not?

Suppose you were asked to investigate which predictors explain the number of minutes that 10- to18-year-old students spend on Twitter. To do so, you build a linear regression model with Twitter usage (Y) measured as the number of minutes per week. The four predictors you include in the model are Height, Weight, Grade Level, and Age of each student. You build four simple linear regression models with Y regressed separately on each predictor, and each predictor is statistically significant. Then you build a multiple linear regression model with Y regressed on all four predictors, but only one predictor, Age, is statistically significant, and the others are not. What is likely going on among the four predictors? If you include more than one of these predictors in the model, what are some problems that can result?

After building a regression model and performing residual diagnostics, you notice that the errors show severe departures from normality and appear to have nonconstant variance. What steps would you take in this case to resolve the errors? If the problems are not corrected after all steps are taken, what does that imply about the modeling approach you are taking? Explain in detail.