Instructor: Jing Gong

Weekly Question #8 (Due Friday, November 13)

Leave your response as a comment on this post by the beginning of class on November 13, 2015. Remember, it only needs to be three or four sentences. For these weekly questions, I’m mainly interested in your opinions, not so much particular “facts” from the class! If you sign in using your AccessNet ID and password you won’t have to fill in the name, email and captcha fields when you leave your comment.

Answer one of the two questions below (not both):

  1. Name and describe a business question that you could answer using a decision tree. What data would you collect to perform the analysis? Don’t use an example we’ve covered in class.
  2. What advice would you give someone regarding how to select the right predictor variables for a decision tree analysis?

38 Responses to Weekly Question #8 (Due Friday, November 13)

  • A business question that you could answer using a decision tree is: Is this consumer likely to purchase from this store again? The data necessary to collect varies from how recently has the consumer purchased, how frequently does this consumer purchase, and when the consumer does purchase, how much do they spend? Once these factors are taken into account, the business question will instill a more accurate answer.

  • 1. One business question i think a decision tree could answer would be how to market certain products. For example, If i ran a sporting goods store and I was wondering where I should market a brand new type of baseball bat I can use old data to decide which gender, age, price range, and demographic would be more likely to buy this new bat. I can then put the amount of advertisement accordingly.

  • What percentage of college graduates are more or less likely to pay back student debt based on Major and in what time frame?
    We can look at colleges individually, identify college graduates, identify the chosen major, how much the student debt was, what is the debt now, paid or not paid in full, we can look at geographical region they live, we can look at gender to separate males from the females, identify graduation year, identify when first job begun, identify how long it took to make the first payment.
    Be aware of what variables that we are using because too much data may throw off the probability distribution and create a problem. We want to select specific variables that will set the perimeters that will get us the best results. If we are looking at student debt payback according to major, the student ID number will not be as important as the income or current debt. We will want to include student major to help identify the major that was recognized by each of the students.

  • Kathryn L Martincic will present on Wednesday. Zachariah Thomas and Jacky Wong will present on Friday.

    • A business question that could use a decision tree is how likely a business student will get a job after graduation. The variables could be employment history, their grade point average, and their previous internships. A student with a high GPA and relevant internship experience would have a better chance of getting a job after graduation then a student who doesn’t have all those criteria. The business school could distribute this data to potential students to increase there enrollment.

  • I used the example of an insurance company choosing whether or not they should choose to insure or not to insure someone for auto personal coverage. Criteria that could be used is age of the car, location, if the person had an at-fault accident within the last 5 years and whether or not they reported a claim in the last 5 years.

  • A business question that you can answer by using a decision tree is in the world of accounting, an accountant has to figure out if a company or individual is eligible for any tax breaks. To perform this analysis, the accountant needs to collect the client’s proper forms, records, and the rules and regulations for applying a tax break.

  • A business question that we could answer from using a decision tree is whether or not to purchase software to benefit the business (versus making/ coding it in-house). The data I would collect to perform the analysis is if the purchased software fills the requirements of the business or if it does not. The other alternative would be if coding it in-house works or does not work. Attached to these splits, would be the probabilities and the amount of dollars it would cost the business. Using these similar branches and interpreting the factors, predictors and (default/no default) we can come to a proper conclusion for the business.

  • A decision tree could help a business answer whether to buy a new truck or keep old truck 10 years for their deliveries. The business want to reduce cost on delivery costs. The variables the business need to consider including price of new truck, the annual payment to pay off a new loan for new truck, the useful of both trucks, the market value of the old truck, miles per gallon for both trucks, number of deliveries annual, deperiacation on both trucks, annual maintenance costs, insurance, and the risk of the old truck breaking down.

  • A business question I would use for a decision tree is, “What characteristics of a hospital patient makes him/her a high risk individual?” I would include important variables such as age, gender, weight, number of medications, insured or not insured, previous injuries or illnesses, family health history, etc. Through the decision tree, we would be able see what characteristics of a person’s health would consider them a low-risk or high-risk individual to the hospital.

  • 2. What advice would you give someone regarding how to select the right predictor variables for a decision tree analysis?

    When selecting predictor variables for a decision tree analysis, first determine what problem you need to solve. What behavior do you want to examine? Consumer buying habits? Web content clicks? College aptitude? Then refer to any data you have that could help you answer that problem. For instance, consumer buying habits can ultimately lead to purchases, which are financial transactions. For this decision tree, select predictors that will help you determine consumers’ financial status and how much disposable income they have. But remember… A data tree analysis is only as good as your data. If you have minimal, incomplete, and/or inaccurate data, your decision tree will be less effective.

  • One question that could be answered using a decision tree is, “Can a customer at a dealership get approved to purchase the car they want”? I would collect the customer’s credit score, employment history, price of the car, and down-payment.

  • A business question that could be resolved with a decision tree would be the decision of whether to hire a prospective employee. Data collected in this analysis would be provided by the candidate as well as a human resources representative. By assessing an individuals qualifications, prior experience, age, and criminal history a company could accurately assess its potential for turnover as well as its return on investment in that employee on a short and long term basis.

  • A question for the residential building business could be, What process do you need to go through to get approved by the township to build a house? The data you could collect is loan amount, material costs, and a housing model and present it to the board. These measures are needed to get approved or unapproved to start the building process.

  • One way that a decision tree can be used is to determine how to look at very complex information simpler for the reader. For example the United States Army uses analytics to try to correctly predict when and where to deploy troops. Using a decision tres they can break down when troops should deploy based on elevation, number of known hostiles, size of the unit, temperature in the area etc and determine the likelihood of when to send the troop.

  • One example of a business question that could be answered using a decision tree is releasing a new product. To perform this analysis, we would need to collect data on competitors products or potential product releases by competitors, market research about consumers, potential cannibalization and potential pricing strategies. By collecting this data and using a decision tree, the company could decide whether or not to release a new product, and if they release a new product they could decide on pricing and other aspects of product release.

  • 1. I think that decision trees could probably be used to determine the amount of risk that a business faces in a typical situation and how they could handle that risk. An example could be if a business is managing the risk that a shipment could be lost and they have to decide how likely a loss could be and how much they would want to invest in order to manage this risk.

    2. I would suggest that someone use a predictor that shows a strong correlation to the variable that you are trying to measure.

  • We can use a decision tree to analyze financial risk. The root node can be INCOME, for income which is from $0 -$15k, put it in High Risk. For$15-35k, direct it into the child node called Credit History. For income that is over $35k, direct it into Credit History. Under the first Credit History child node, if the history is unknown direct it into Debt, if it’s bad direct it into High Risk. If it’s good direct it into Moderate Risk. For the second Credit History child node, unknown—Low Risk, Bad—Mdoerate Risk, Good—-Low Risk. For the Debt child node under the first Credit History child node, should categorize it into High debt and Low debt. If it’s low debt, put a leaf note of High Risk under it. If it Low debt, put the leaf node called Moderate Risk under it.

    The data should be collected are credit scores, annual income, debt amount (if any), debt statements, etc.

  • Actuaries use a more complex version of a decision tree to calculate the probability of a person dying when pricing their life insurance. They would include data such as is the person a smoker, the person’s weight, the car the person drives, where the person lives and if there are people dependent on the person’s income into their model. The model will calculate a probability and actuaries can use this number to price a life insurance accordingly.

  • A football franchise could use a decision tree to decide whether they should run the ball or throw the ball on a certain play. It could collect data on how many yards teams average when running or throwing in a certain situation. The team could also collect data on the percentage of touchdowns scored running and throwing the ball against specific teams.

  • A business question that could be answered using a decision tree is “How likely is a student to enroll at an Ivy League University”. Some of the data that you could gather is things like race, gender, GPA, High School, Location of High School, Where you live, family income, personal income, and scholarship opportunities, financial aid, and more!

  • From my personal experience as a Forex trader, I can use the decision tree to decide whether to buy or sell a currency set based on different outcomes. I would start out with the currency pair as the root. For the different data I would collect are the different graphs used to record the values, the different intervals that go along with the graphs, and the analytic method I use to decipher each graph. The graphs can be used to show different trend lines and how well the pair is doing. The intervals will show what happened at the past intervals and we can use that to determine what will happened in the future. The different analytic methods determines how high or how low the pair will go and where it will happen.

  • An email management decision tree might begin with a box labeled “Receive new message.” From that, one branch leading off might lead to “Requires immediate response.” From there, a “Yes” box leads to a single decision: “Respond.” A “No” box leads to “Will take less than three minutes to answer” or “Will take more than three minutes to answer.” From the first box, a box leads to “Respond” and from the second box, a branch leads to “Mark as task and assign priority.” The branches might converge after that to “Email responded to? File or delete message.”

  • A business decision that can be made using a decision tree is whether or not to send a customer some coupons for your store in the mail. Basically, a business can decide whether or not the customer is likely a reoccurring customer, if the customer has given personal information to them such as an email address, or whether or not the customer is likely to spend enough money to qualify to use the coupon or to benefit the business. The advice that I would give to the business who is looking to increase sales via coupon distribution would be to ensure that the predictor values that they use maximize the distinct outcomes by ensuring that they bring in the correct clientele who will maximize revenue and sales and also make sure that the predictor values that they use separate the outcomes as to whether or not to invite a specific customer.

  • A business question that could be answered using decision tree could be whether to expand a company by opening a new location. You would need to find out are the state of the economy in general and in the area you are looking to put a location, the demand for the products offered in that area, and other success of existing locations. However, the data collected should not be limited to those and should become more specific depending on the type of company, their future goals, and the current state of the company compared to the state of the market.

  • Apple’s decision on to releasing new iPhone(iPhone 7) or developing a new cheaper updated version of iPhone(iPhone 6c). The data that needed to be collected is age(phone size like if kids use it, make phone smaller), demographic(which group of people use it) , price(how much is people willing to pay), color(will increase the probability of people buying iPhone because they want to match their favorite), and durability(how much people use their phone).

  • A business question that could be answered with a decision tree is how likely a homeowner will pay off their mortgage in a certain number of years. Collected data could include household income, loans, expenses, credit score, and employment history. This could help the homeowner to see how long they will take to pay off a mortgage as well as helping the bank with determining loans or refinancing.

  • A good business question that could be answered using a decision tree is for Financial Risk. Using a decision tree you can tell if someone is high or low risk, based off of your criteria. Then knowing their level of risk will be able to answer the question if you should or shouldn’t insure this person or client.

  • In business, the decision to whether there is financial risk in allowing a loan can be answered using a decision tree. You would be able to split tree by income amounts and assess the risk through the credit history and whether the risk would be high, low, or medium. You would collect data related to income and credit history which would allow you to determine the risk involved.

  • You should select the splits on the predictor variables that are used to predict membership in the classes of the dependent variables for the cases or objects in the analysis.Starting with the split at the root node, and continuing with splits of resulting child nodes until splitting stops

  • A business question that could be answered using a decision tree could be “how likely is X customer to purchase Y product in the following year”. Using the decision tree, you could break down the different potential options in to a series of questions and you could simply follow along the tree answering yes/no. Eventually this would lead to the answer to your proposed business question.

  • One business decision question that we could answer using a decision tree is whether or not someone gets approved for a loan. You could collect credit history, rent, and income from someone qualifying. If someone doesn’t get approved someone could try adding a cosigner to the loan and then go through the questions again. Loans work the same way as applying for a store credit card where your credit and credit history matters a lot when considering you as an applicant.

  • Using a decision tree, I can answer whether a person is eligible to get a rental car from Budget or not. I would use age , whether you or using a debit or credit card, whether or not you have a valid id, and whether you have a third form of identification. The age requirement is 21 years old in Philadelphia. If a person is 21 years old with a valid license, acceptable credit card, and a third form of identification the person will be eligible to rent a car.

  • Decision Trees are very useful tools to have when making business decisions. An example of a business question that a decision tree could help answer could be “Which product has the least demand at x location”. You could pull data about the product quantities sold in different time periods to find the least sought after item, or by revenue to find which product brought in the least amount of money in a period.

  • Selecting the right predictor variables is crucial to creating a successful decision tree. The predictor variables must provide insight that answers the questions that are being sought. All factors that effect these questions should be accounted for by a predictor variable.

  • Decision trees are used can assist executives in making strategic decisions. An executive uses a decision tree to decide wether or not to undertake a certain project that would be quite expensive. An example of that would be If a company wants to decide if they should pay all that money to air a Super-Bowl commercial. The company wants to know if their marketing campaign will be a success or failure. SO they use a decision tree for this. Some factors to consider are the appeal the commercial will have to the audience and how the economic state.

  • A business question that could be answered by using a decision tree is if whether or not a company such as Franklin Mint Federal Credit Union should launch a new loan rate as a special. The company should gather data regarding: state of economy (recession or normal state) consumer demand (how likely are potential consumers to purchase a new home or etc), income of target consumers, etc. These are some of the factors that could potentially impact the success of the new loan rate being released.

  • Large corporations and government agencies often struggle with how to handle uncertain future scenarios. But, the constantly changing business climate requires executives to plan for a variety of outcomes to remain competitive. Decision Trees can help with this, one kind of question is finding out what kind of new product would be successful in the market place. It takes variables from multiple sources so the executives can make the right decision.

    tue97209, Robert Gallagher

  • An example of a business question that could be solved using a decision tree would be whether or not to accept or reject a pool of applicants for a job interview. That could discriminate applicants that, for example, did or did not graduate college. They could also discriminate by total work experience, periods of unemployment etc. This would save time for reviewing applications, so a company has a shorter list of who to possibly call in for an interview.


    Decision Tree is pretty useful when you are thinking about investing in a new business strategy. We have to consider the risk level, is it risky or not? If it’s not much risky, we also should think about what kind of improvement this new strategy does, is it efficiency? Is it get access to the market? And so on. If all of the answer are yes, then it worth us to invest in. If there is any NO answer, we have to re-evaluate it and improve it.

Leave a Reply

Your email address will not be published. Required fields are marked *

Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 16 other subscribers