What is Predictive Modeling
So, what is predictive modeling? I have designed the course to teach you just that and more! By the end of the course, you will understand the process of predictive modeling, so you’ll know how to create a predictive model by yourself. I will of course cover much more than the process and the building of a model, and you can see the whole curriculum near the bottom of the article (course outline).
We will work with a data set where the goal is to predict a customers Loan_Status. That is, we want to predict whether a client will receive a loan or not based on the information (variables) we have about them. For example, we might have their age, their income, their job title, the number of dependents they have, their gender, etc. To make this prediction for the future, we will work with historical data (this is data from the past that has the outcome, ie. the data-set includes the variable for whether someone received the loan or not). We will then use this data-set, with a specific algorithm, to train on this data set. The training process is about the computer learning, from historical data, how the predictors (ie. age, income, etc) are related to the outcome (ie. if the person got the loan or not).
In the end, we will judge how our trained model performed by how accurately it predicts the outcome (Loan_Status) when we get it to make predictions on unseen or new data.
Is Predictive Analytics the same thing?
Predictive analytics includes a variety of statistical techniques from data mining, predictive modeling, and machine learning, that analyze current and historical facts to make predictions about future or otherwise unknown events. source: https://en.wikipedia.org/wiki/Predictive_analytics
So predictive analytics is simply broader. So what is predictive modeling? It’s just about utilizing a more narrow number of statistical techniques to make predictions.
Predictive Analytics in Financial Services?
Predictive analytics are quite common in the financial services and beyond. For instance, I have seen it utilized in vehicle insurance. The purpose is to assign a risk of incident to policyholders from information obtained from those same policyholders. For example, I would use a policyholders information, like their age, the number of years they have driven, the number of accidents they have had thus far (ie a bunch of predictor variables), to predict the chances or risk that they will have an incident with their vehicle in the near future.
The financial services are not the only industry relying on predictive modeling. It is common for hospitals to use patient records to determine which patients have the highest risk of being re-admitted. Imagine how many more lives could be saved by being more prepared? In sales, it is widely used to decipher the churn probability for each customer.
Now, predictive modeling is an attractive option because it brings a ton of value to any organization. However, it is also on the higher end of the complexity spectrum when compared to some other business analytic options. The complexity in utilizing predictive modeling largely manifests in the number of steps required and the expertise required for some of the steps.
I designed this course as a way of reducing and/or eliminating the expertise required and the time it would take to go through all the steps. I think this course is useful for anyone that wants to save time and money.
Why SAS Predictive Modeling?
SAS is the programming language I have used for the course. While it is not necessary to be familiar with SAS to get a lot out of the course, if you have the time, I would suggest learning SAS here: SAS Online Tutor for Base Programming
Have a read of that whole page. You will begin to appreciate why I use SAS, and why you should learn SAS, as opposed to other programming languages. I have been listening to people talk about SAS dying out in-favor of Python and R for a long time. If you already work for a Fortune 100 company, you probably already know why this is untrue and why understanding how to code in SAS is so crucial.
Now, speaking of Fortune 100 companies, if you are currently working at one or intend to work for one in the near future, they will likely be utilizing SAS Enterprise Guide and/or SAS Enterprise Miner. For predictive modeling specifically, you will likely be using SAS Enterprise Miner. However, this does not mean that this course, or my SAS programming course, will not be useful. In fact, you can utilize SAS code in both of these software. I can guarantee that it will give you a distinct advantage over your colleagues. You will not be restricted to the point and click interface like most of your colleagues. I have heard from numerous students how taking my SAS course has accelerated their career. So, I don’t want you to wait.
What is Predictive Modeling, Step by Step
The course answers the question “what is predictive modeling” through a step by step process. For example, one of the first video lessons is about understanding how modeling is used in the business context, while the remainder of the course lessons are really about breaking down the modeling process. Below is the course outline:
Understand the challenges in the analytics world
Understand how modeling is applied in the business context.
The machine learning algorithm utilized for the course is logistic regression. So you will gain intuition how logistic regression algorithm works and when it should be used. The algorithm used depends on a few factors, which I discuss in one of the opening lessons. There will be situations where you would use support vector machines, decision trees, random forest, etc.
Learn how to perform a data audit.
How to perform univariate and bivariate analysis.
Techniques to deal with missing data. (ie mean, mode, median and multiple imputation)
Oversampling and how to adjust for oversampling.
Dimension Reduction for your Categorical Inputs.
Subset selection (utilizing automatic techniques for variable selection or feature selection).
How to analyze multicollinearity in your features.
Partition your Data, Train a Model, and Make Predictions on Unseen Data.
Analyze Classifier Performance (ROC curves, Optimal cutoffs, K-S).
You will be provided with the full SAS code.
Here is a free preview of one of the lessons. In this video I discuss Analytics challenges.
Thanks for viewing the free preview. Do not wait any longer, register for my course NOW!