Assignment essay paper
Assignment
ORDER A PLAGIARISM FREE PAPER NOW
Question/Issue
I’m slightly stumped on how to begin this project.
Answer/Response
I recommend you get your data in Tableau and get Assignment everything set up correctly (think defaults and other things like we learned in class). Then, start exploring the data by creating charts like we did in class. As you are doing this, remember that your end goal is to tell a story about attrition, so many of your charts will probably encompass that somehow.
I’ll also throw this out there. I tend to be very methodical about how I explore data. My process for this project would look something like this:
1. Load the data.
2. Prepare the data. Assignment
1. Look at the raw data and compare it to the description. Is it what you expect it to be? If not, try to figure out why is there a difference and if you can fix it (e.g., with a calculated field).
2. Set appropriate default properties (formatting and aggregation) for each measure. After doing this, look at the raw data, again. Is it what you expect it to be, and if not, why and can you fix it?
3. Explore the data.
1. Create charts that describe each variable by itself.
1. Create bar charts for each dimension (the categories on one axis and the count of how many records have that value for that variable in the other axis).
1. How many different categories do each of the dimensions have?Assignment
2. What are the categories? Are they coded, and do you know what the coding represents?
3. Describe the bar chart. Do you notice any pattern?
4. If the dimension has an inherent order, set the default sort order to be based on the inherent order. If the data is nominal, sort based on frequency.
2. Create histograms and/or box and whisker plots(disaggregated) for each numeric variable.
1. Are the distributions what I expect based on the knowledge I have about the subject?
2. Are there any outliers?
3. ** Note that we didn’t cover box and whisker plots in class, but some of you may be more comfortable with box and whisker plots from other classes or experiences. Assignment If you use box and whisker plots, make sure you disaggregate the data.
2. Create charts for each combination of two variables (e.g., scatterplots for every pair of numeric variables, line charts for every combination of date and numeric variables, etc.)
1. Do the patterns and trends make sense? Assignment
2. ** This can include text or highlight tables.
3. ** For the culminating assignment, it would be especially helpful to somehow combine attrition with all of the other variables.
3. Create charts based on combinations of three variables.
1. Do the patterns and trends make sense?
2. ** This can include text or highlight tables.
3. ** For the culminating assignment, it would be especially helpful to somehow combine attrition with the other variables.
4. Think about the data and ask yourself questions.
1. Are there any questions that come to your mind when you think about the variables you have? Try to answer them by creating charts.
2. What assumptions are you making about the data? Check to make sure those assumptions hold. If they don’t, try to figure out from the data and information you have what is happening with the data instead.
3. ** At this point, I’d check to make sure I have the minimum number of calculated fields needed, and if not, I’d think through if there are any I can make to help with my analysis and create them.
4. Create visualizations in Tableau that show interesting characteristics of your data that help answer the business question. These visualizations may come from the visualizations you created in the steps above.
1. “When we do exploratory analysis, it’s like hunting for pearls in oysters. We might have to open 100 oysters (test 100 different hypotheses or look at the data in 100 different ways) to find perhaps two pearls.” (Knaflic 2015, Assignment Exploratory vs. explanatory analysis).
2. ** Your exploratory analysis is what you did in step 3. These visualizations should be the “pearls” you find. These are what I’d recommend including in your dashboard. Don’t worry if I agree if what you find is interesting. You can provide a justification for your point when you create your presentation.
3. ** Remember: Just because you can build a visualization (specifically a complicated one), doesn’t mean you should.
I then use my visualizations from step 4 to create my dashboard. That said, this is how I would tackle this project. Others may have a different strategy that works for them.
Question
I mostly understand updating the metrics to the correct number format, but I get confused on which aggregation to select since I sometimes don’t want to do any calculations. For instance, for monthly income, I want it to stay the monthly income number for each employee number instead of totaling.
Answer
You will want to set your default properties correctly for data aggregation for every field you use. You can choose to disaggregate the data (which isn’t affected by your default aggregation statistic) like we did in Module 4 when we built the scatterplot (see the Disaggregating for Scatterplots lecture video). If you are thinking about disaggregating your data, I recommend considering how best to present the information you want to convey. There are some specific instances when disaggregation is needed, but if you are disaggregating everything (which it doesn’t appear you are at this point), I would ask “Is that the best way to summarize your findings to the CEO?”.
The question then becomes “How do I know which default aggregation statistic to use?”. Unless you are doing something very specific and on purpose, you will likely be using sum or average for the assignment. It is important to choose the correct of those two. For example, it doesn’t make sense to add percentages because the result isn’t meaningful. However, the average percent is more meaningful in part because it can be viewed on a scale we are familiar (e.g., on average x percent of employees left their job). This was discussed briefly in one of the lectures. Another way to think about whether sum or average is appropriate is to determine if the number of values you are aggregating can cause you to misinterpret results. For example, let’s assume that very few people left their job. Now consider one of the measures; let’s use income. If we compare the sum of income between employees who didn’t leave (a large group of people) and employees who did leave (a small group of people), what insight does that provide? Mostly, it only tells us that there were many more people who didn’t leave their job than those who did, but it doesn’t really speak well about income differences. However, if we use average income when comparing these two groups, we can see the difference in average income between these two groups which provides insight about income differences. Make a couple of graphs to try this out to get a better visual. Large differences in group sizes really make this apparent; however, small differences in group sizes are also affected. Assignment
Question
I am having a hard time making any of my graphs provide meaning with the measures given. I understood our examples and test material much better when we were working with dates and measures like Sales or profits. I have tried similar graphs with the data points for this final project, but most are looking very flat and repetitive with most of the data points being between a 1-4 rating. I am trying to use averages, but again it’s not giving me a lot of change to work with. I have ideas I want to create and am having a rough time making them look distinct from each other since almost every average is between 2 and 3.
Do you have any advice to help point me in the right direction?
I guess my biggest issue is that I don’t know how to create visualizations using this data since it’s a rating scale after we’ve practiced so much with sales. Like I said I’ve done a little work, but I keep coming back to the same type of visualizations and it’s all looking too similar.
Answer
I have a few suggestions for you.
• Remember that the problem you are trying to solve is about attrition. While I wouldn’t say that attrition will have to factor somehow into every viz, it will probably be in most somehow because that is what you are investigating.
• Regarding the small differences, I wouldn’t be concerned that the differences are small when you are looking at the survey results measures. The questions I recommend you think about:
o If the differences are small, is this likely a major reason why people are leaving?
o Are there other factors where there are larger differences between the folks who are leaving and the folks who are staying? For example, maybe income and other benefits would factor into whether someone stays. Then start thinking of combinations of reasons why some folks leave while other folks stay. If someone is close to retirement, they may be more likely to stay than others who would leave for the same reason.
• I also wouldn’t worry that your graphs all look the same (i.e., uses type of graph/table). You want to use the viz that is best suited for the data you are trying to display and the story you are trying to tell. Sometimes that means you end up with a bunch of bar graphs. I recommend you think about the visualizations we used most in class and in our dashboards.Assignment
I see this assignment as a two-step process. First, you need to explore the data to find the story you want to convey. Second, you need to decide what the best way is to present the information. Here’s quote from a data visualization book that I feel conveys this well.
“When we do exploratory analysis, it’s like hunting for pearls in oysters. We might have to open 100 oysters (test 100 different hypotheses or look at the data in 100 different ways) to find perhaps two pearls. When we’re at the point of communicating our analysis to our audience, we really want to be in the explanatory space, meaning you have a specific thing you want to explain, a specific story you want to tell—probably about those two pearls.” From Storytelling with Data: A Data Visualization Guide for Business Professionals by Cole Nussbaumer Knaflic
I don’t think you’ll have to test 100 different hypotheses to find a reason for the attrition problem, but you should keep looking. I think you’ll see some instances where there are more obvious differences than what you’ve found already. Remember, you are trying to find reasons why people are leaving, so in your early exploring, I would keep the Attrition variable in your visualizations, so you can find major differences in the people who are leaving and those who are staying. You should also think about the numbers of people leaving v. staying. If only 1 out of 100 people left in a particular circumstance and you say a difference in the people who left and the people who stayed that’s one thing. If 50 out of 100 people left in a particular circumstance, it presents a different picture.
This is an open-ended project, so I’m trying to guide you loosely. If you need more help, please let me know.
Question
So far, I’ve been able to develop highlight tables, scatterplots, and some bar charts. I would love to analyze the data in additional ways but am getting stuck here. For instance, there is no location data for a map. Should I try to create a wider variety of graphs or more complex/multi-faceted visuals? I know some of this will come once I implement the dashboard.
Answer
We do not talk about visualization theory much in this class. One key aspect of visualization is that you want to use the chart type that best communicates the information you want to convey. Often that means you end up using the same chart type over and over again, and that’s OK. A bar graph is the best way to compare values. It’s easy to see which category had the highest or the lowest value, especially if you order the bars in ascending or descending order. That said, I wouldn’t worry about the variety of charts you have in your dashboard. I would focus on making sure that you clearly convey your findings by using the most appropriate chart type.
