Learn logistic regression modeling in 1 week

Name: Learn logistic regression modeling in 1 week
Rating: 3.3 (5 reviews)

A hands-on, project-based training to help you quickly learn predictive modeling skills through real-world project work.

Created byPredictive Analytics

Last updated 8/2014

English

English [Auto],

What you'll learn

Learn how to develop a logistic regression model from the beginning to the end
Learn how to interpret modeling results
Learn how to present modeling results to others

Course content

8 sections • 26 lectures • 2h 18m total length

Course Objectives, Outline, and Prerequsites6:50
In this section, we will explain why predictive analytics is an important skill, what you will learn from this class, what is so unique about this training class. We will also provide course outlines and prerequisites for taking this class.

Project Overview4:00
In this section, we will provide modeling project overview. This session will cover:

Business background, business objective and the challenges.

Data to used in the project

High-level analytical solution

Quick summary of key learning

Web link for data and document download for the project:

http://www.sigkdd.org/kddcup/index.php?section=1998&method=info
Project Overview - Part 24:18

Logistic Regression Modeling Methodology Overview4:42
In this section, we will provide high-level overview of modeling methodology. First, we will describe two different types of logistic regression model. Then we will discuss modeling equation for binary logistic regression. Next we will explain how variables and their coefficients are being estimated using statistical approach. We will end the section with a summary.

Modeling Process Overview3:06
In this section, we will first walk you through our 6-step modeling process. Then we will dive into each of the modeling steps. In each step, we will do extensive demo to show you how to use or develop SAS program. We will run the SAS program and provide explanation on SAS outputs. For you to get the best out of this section, we recommend that you open your SAS program and try to execute the same program along with us.
Modeling Step 1A - Data Loading6:00
Step-by-step demo on how to load a text file into SAS
Modeling Step 1B - Variable Labeling5:33
Step-by-step demo on how to apply labels to SAS variables
Modeling Step 1C - Data QC5:09
Step-by-step demo on how to performance data checking and create QC report
Modeling Step 1C - Data QC - Part 26:14
Apply data quality checks using proc freq and proc means to assess dependent variable distribution. Transform date variables into durations and flag 99 percent missing and exact-same-value variables.
Modeling Step 1D - Data Transformation3:20
Step-by-step demo on how to do data transformation
Modeling Step 1D - Data Transformation - Part 26:10
Transform date observables into duration variables in sas using arrays and do loops, then create a 50/50 training–validation split with a random sample and verify distributions with proc freq.
Modeling Step 2 - Select Variables7:07
In this section, you will learn how to develop / use SAS tool to select top 50 to 100 most important variables out of up-to-hundreds of potential predictors. This helps to focus our efforts on important variables for further deep dive analysis.
Modeling Step 3 - Deep Dive Analysis5:00
The objective for this section is for you to learn how to use a SAS visual tool to discover the relationship pattern between predictors and target variable. We will use this relationship pattern to help us transform variable in next section.
Modeling Step 3 - Deep Dive Analysis - Part 26:00
Demonstrates a SAS macro workflow for logistic regression: define input data and dependent variable, bin predictors into 10 groups, and build summary tables to inspect relationships.
Modeling Step 4 - Transform Variables3:29
In this section, you will learn how to use bivariate plots for variable transformations, how to impute missing values, how to handle extreme values, and how to perform non-linear transformations. At the end of the class, you will be able to perform sophisticated variable transformations.
Modeling Step 4 - Transform Variables - Part 26:50
Apply a bearable transformation macro to predictors, set imputation, raw, and KEF values per predictor, and use median imputation with a cap at 30.
Modeling Step 4 - Transform Variables - Part 35:25
Modeling Step 5 - Develop Model6:46
In this section, you will learn how to develop logistic regression model, how to run model diagnostics, how to create model scoring equation, how to score whole modeling data, and how to create model performance chart. At the end of the class, you will be able to develop a logistics regression model independently.
Modeling Step 5 - Develop Model - Part 23:50
Develop a SAS-based stepwise modeling workflow to generate a model scoring equation from coefficient estimates, including data step, transpose, cross-product view, for training and validation.
Modeling Step 5 - Develop Model - Part 34:07
Modeling Step 6 - Validate Model6:45
In this section, you will learn how to prepare validation data, how to score model on validation data, and how to validate model performance. At the end of the class, you will be able to evaluate model performance on validation sample.
Modeling Step 6 - Validate Model - Part 26:42

Requirements

Microsoft Excel and Powerpoint
SAS Software with Base/Stat/Graph Components

Description

This is a hands-on training class for those of you who want to learn or improve predictive modeling skills. We will focus on developing and validating a logistic regression model in this class.

We will walk you through our 6-step modeling process step-by-step. Together, we will solve a real-world modeling project with real-world data. We will provide extensive demo on SAS code development, running SAS codes, and explaining SAS outputs. We will also provide you all the SAS codes used in this course.

At the end of the class, you should be able to develop a logistic regression model independently.

Who this course is for:

Working professionals who want to learn predictive modeling skills
Graduates or undergraduates who want to explore their career opportunities in advanced analytics

Learn logistic regression modeling in 1 week

What you'll learn

Explore related topics

Course content

Introduction1 lecture • 7min

Overview1 lecture • 7min

Project2 lectures • 8min

Data1 lecture • 7min

Methodology1 lecture • 5min

Modeling18 lectures • 1hr 38min

Results1 lecture • 7min

Appendix. Downloadable SAS Codes Used in the Section 61 lecture • 1min

Requirements

Description

Who this course is for: