WGU C207 Data-Driven Decision Making Exam Guide, Answered (Verified Solution)

WGU C207 Data-Driven Decision Making Exam Guide, Answered (Verified Solution) Define analytics Turning information into knowledge and developing fact based strategies to gain a competitive edge. Descriptive/Diagnostic analytics Encompasses the set of techniques that describes what has happened in the past. Predictive analytics Consists of techniques that use models constructed from past data to predict the future or ascertain the impact of one variable on another. Prescriptive analytics Decision models that indicate the best course of action to take. Three characteristic of big data 1. Structured and unstructured large volumes 2. Analytics is used to discover and communicate meaningful patterns 3. Analysis of big data requires a system of organization Davenport-Kim Three-Stage Model* A decision-making model developed by Thomas Davenport and Jinho Kim that consists of three stages: 1. Framing the problem, (why are you trying to do this and what do we already know) 2. Solving the problem (Choosing the model, collecting the data, analyzing the model) 3. Communicating results (Communicating and acting on the results) Categorical or discrete data variables Represents types of data that can be divided into groups or categories. Example: race, sex, age group, and educational level, happy, etc. There are two types of categorical data Nominal and Ordinal Nominal data* Identifies, groups, or categories only. Data cannot be arranged in an ordering scheme. They are used to name or label a series of values. Examples, names, pass/fail, gender, race, the color of the eye Data can not be ordered and makes no sense to calculate means or standard deviations off this. Ordinal data* Data is placed into some order by some quality. They are usually non-numeric captions like happiness or discomfort. It is the ORDER that matters. They provide good information about the order of values, like a customer satisfaction survey. Examples: top ten best cities to live in or top 10 college football teams. (ordinal-order matters) Numerical or continuous variables They can accept any value and can measure nearly anything. Height, weight, BP, etc... Two types of numerical data Interval and Ratio Interval data* It has an order to it and has equal intervals apart. It gives us the order of values with the ability to quantify the difference between each one. Differences between values can be found and meaningful, but they DO NOT have a true 0. ( no temperature, no true 0 time, or no true 0 dress size) These can be added or subtracted but not multiplied (interval means space in between which is the important thing to remember). Ratio data* Data similar to interval data but does have an absolute 0. Ratios are meaningful. (Length, Width, Weight, Distance, age, income, stock prices, repeat customers...) Data can be added, subtracted, multiplied, or divided. Qualitative research Data not characterized by numbers, usually textual, visual, or oral responses. Examples: Hair color, car colors, letter grades, types of coins in a jar. ANYTHING THAT DOES NOT HAVE A NUMER IN IT! Quantitative research Gives you numerical data. Three elements to a quantitative study design 1. Units, this is the subject or object that is being observed. (Example: Students are broken up into three different groups) 2. Treatments, the procedures being applied to each subject. (Examples: which learning is best, lecture, simulation, or no treatment. Each unit gets a different treatment in this example) 3. Responses, the effect of the experimental design. (What were the results of each group, who got the best grades based off the different treatments.) Data managements responsibility Clean and organize data sets, make sure clean good data is provided. 5 questions asked about data management 1. Is the data relevant to your business 2. Is the data complete, do you have all the data 3. Is the data accurate 4. Is the data available and assessable 5. Is the data timely and up to date Do all data measurements have some degree of error? Yes 4 common errors in data 1. Random errors, caused by unknown and unpredictable errors. The environmental conditions or the measuring instruments cause these errors. These can be minimized by increasing the sample size. 2. Systematic errors, error in the data or measuring instrument but it is a constant error. Like a measuring instrument not returning to zero. 3. Omission errors, when something has been left out or when an action has not been taken. 4. Outlier errors, observation points that are really far from other observations. You need to figure out if it is wrong or if it is actually correct and should be included in the data set. Skewness How much the data "leans" to one side. What is measurement bias? The bias that occurs from not selecting a true random sample. When the sample is not representative of the population, the sample test must be sufficiently random. What is Information bias? When the respondent or interviewer has an agenda and is not truly unbiased. What is response bias? When respondent says what they believe the questioner wants to hear What is conscious bias? When the surveyor is actively seeking a certain response. Maybe you only choose to ask women, or children a question. That is consciously choosing to be biased because of a preference. What is a triple blind study? When the data Analyzer, data gather and participant are not told What are Random errors? 1. Random errors, caused by unknown and unpredictable errors. The environmental conditions or measuring instruments cause these errors. These can be minimized by increasing the sample size. What are systematic errors? Systematic errors, error in the data or measuring instrument but it is a constant error. Like a measuring instrument not returning to zero. For example: a plastic tape measure gets stretched out over time. The error repeats itself. What are omission errors? Omission errors, when something has been left out, there is missing data, or when an action has not been taken. What are outlier errors? Outlier errors, observation points that are really far from other observations. You need to figure out if it is wrong or if it is actually correct and should be included in the data set. T-test* Is used to compare the average between two groups. Keyword comparison between 2 groups and AVERAGE. Or, to compare the average to a standard (1 sample T-test. Example: Pizza delivery times between car and bicycle. ANOVA* Same as at T-test but can compare MORE than 2 groups. Used to compare the mean of three or more groups DIFFERENCE IN AVERAGE For Example, Pizza delivery times between, bicycles, cars, and scooters. When using ANOVA the test will tell us that two or more means are statistically different but we are not able to specify which ones they are. Chi-Square* They allow us to compare frequencies between groups. DIFFERENCE IN FREQUENCY Example: Late pizza deliveries, compares the frequency of late deliveries between different shifts. Why use regression? To help predict an outcome. We use variables to help predict an outcome. Linear regression A number predicting a number. Using an actual number A technique using a single independent variable to predict a single dependent variable USES THE WORD PREDICTOR or PREDICTION Do hours studies predict your exam score? How many minutes doe it take to complete an order? Logistic regression* A number predicting a Yes/No outcome only, it is categorical. Example: Is an order completed on time? Yes/no, treated/not treated, etc. How big does a sample size need to be? 30 or more to assume normal distribution. Linear Regression (multiple) When there is more than one predictor variable. Example: Number of pizza slices, and number of toppings are evaluated to see how many minutes it takes to complete an order. Time Series When we are measuring time. A simple regression using time as the independent variable Example: Quarters in a year. Break even analysis When you have covered your cost and your next sale will yield a profit. How many pizzas do you need to sell to make a profit? You need to know: 1. Fixed Costs 2. Variable Costs 3. Price per unit, what you are charging. Cross Over Analysis Determines when one option is better than another option. Tells us the best cost per volume produced. Keyword: VOLUME Bicycles are cheaper unless we deliver more than 50 pizzas per day. You need: Fixed costs for EACH option Variable cost per option Which delivery option is optimal? Secondary data collection methodology When the data already exists, but you need to collect and compile it for research Primary data When you need to collect the data via interviewing, observing, or surveying someone. Then you take the results and translate them into helpful information. Random Sample A sample that fairly represents a population because each member has an equal chance of inclusion Think bingo balls. Responce bias Is when the responder feels persuaded to answer a certain way or that you only have one answer. The question IS NOT persuasive, but you might answer differently if someone else asked you. Example: Your boss asks you, do you support working overtime. Or, do you support neighborhood safety? Well yes, then you will support a 1% tax increase to keep our kids safe. Conscious Bias When the researcher creates a bias in the question phrasing. Lawyers call this leading. "wouldn't you agree" you are being led down a path with the verbiage in the questions. Blinding It removes previous experience or perceptions. Example: a blind tast test. Causality Correlation is not cause!! This is the most misused. Just because things go together that does not mean they cause each other. Example: what time you leave your house every day is highly correlated with the amount of traffic. However, what does what time you leave your house cause the volume of traffic? No, the traffic still exists without you. Multiplication rule The multiplication rule is a way to find