Using Predictive Data Analysis
The sheer variety of sources and types of data that can aid in decision making are almost overwhelming. The key to making good use of the data lies in knowing what specifically to pay attention to, understanding the relationships and variables among the data, and making the right connections.
Experience is essential to knowing and making educated guesses about what requires attention. Familiarity with statistical methods will provide you with a significant advantage over relying on gut instinct alone.
In this course you will learn to identify uncertainty in a business decision, and to choose variables that help reduce uncertainty. By the end of this course, you will have a robust decision model that you can use to make predictions related to your decision. Along the way, you will clarify and enhance your understanding of the factors that influence possible outcomes from the decision.
Who Should Take this Course?
This course is appropriate for anyone from analyst to the SVP with no background in statistics. It is designed for individuals who need to perform analysis to support decision making. The course content draws on examples across all business types.
Participants who complete this course will be able to...
- Determine the degree of uncertainty in your decision and determine the impact of this uncertainty
- Identify data relationships to reduce uncertainty
- Create a regression model that looks at attributes of variables driving the decision
- Refine your regression model to improve its validity
- Create a convincing argument for the validity of your model
- Make a prediction or an estimate using your model
Module 1: Discovering Relationships
- Using Attributes to Create a Useful Model
- Interpreting Correlation Coefficients
- Recognizing Nonlinear Associations
- When and How to Do a Linear Transform
- A Revinate Study Shows Nonlinear Impacts of Engaging with Consumers Identify Variable Relationships
- Analyzing in Your Area of Expertise
- Consider How a Variable Impacts Your Decision
Module 2: Quantifying Impact
- Describing Relationships with a Scatterplot and Best Fit Line
- Obtaining the Equation for a Best Fit Line
- Estimate Slope and Intercept for a Best Fit Line
- Testing for Statistical Significance
- Run a Single Variable Regression on Data
- Using Multiple Regression to Consider Several Relationships Together
- Run a Multiple Variable Regression on Data
- Coding Categorical Independent Variables
- How Available Is Your Data?
- Map Decisions to Outcomes
Module 3: Assessing and Validating Your Model
- Accounting for Variable Interactions
- Addressing Multicollinearity by Reviewing a Correlation Table
- Assess Multicollinearity Using a Correlation Matrix
- Using Residual Plots to Detect Nonlinearities
- Addressing Nonlinearities Using Transforms
- Modeling Interactions between Independent Variables
- Considering Missing Variable Bias
- Generate a Revised Regression Equation
Module 4: Applying the Predictive Analytics Framework
- Using Your Regression Model
- Predictive Analytics Framework Diagram
- Testing Your Model with a Holdout Sample
- Using Logistic Regression to Model Categorical Variables
- Segmenting by Creating Artificial Categories
- Validate Your Model
- Christopher Anderson, Professor, School of Hotel Administration