TensorFlow For Dummies

Chapter 6

Analyzing Data with Statistical Regression

IN THIS CHAPTER

Identifying trends with linear and polynomial regression analysis

Classifying points with logistic regression analysis

Modeling systems with the logistic and softmax functions

Computing loss with log likelihood and cross entropy

Everybody knows that machine learning is a fast-paced, exciting field for clever, future-minded people, and everybody knows that statistics is a boring, stodgy field for people who enjoy Muzak. So newcomers may find it odd to see a chapter on statistical analysis in a book on machine learning.

But machine learning and statistics have a lot in common. In fact, they have the same ultimate goal: to model real-world systems with mathematical relationships. Machine learning relies extensively on statistical methods, and this chapter presents three methods that play critical roles in TensorFlow development: linear regression, polynomial regression, and logistic regression. In addition, the example code in this chapter solidifies the manner in which TensorFlow applications perform training.