Feature Engineering: Machine Learning With TensorFlow On Google Cloud Platform Specialization

132

[editorskit display=”wordcount” before=”Reading Time: ” after=” min”]

Feature Engineering:

Are you interested in knowing how you can improve the accuracy of your machine learning model? Or concerned to know which data columns make the most useful features? In this Feature Engineering Course, Machine Learning Courses by Google Cloud Platform where there will be discussion on elements of good vs. bad features and how you can pre-process and change them for optimal use in your machine learning model.

In this course you will get the facilities to choose hands-on practice and pre-process them within the Google Cloud Platform with interactive labs. The instructors in this course will walk you through the code solutions, which will be made public for your reference as you work on your future data science projects. This Feature Engineering, Machine Learning with TensorFlow on Google Cloud Platform Specialization’s course number 4 can be completed within 1 week in about 11 hours of study.

Syllabus – Feature Engineering

Week 1: Introduction

Want to know how you can improve the accuracy of your ML model? How to know which data columns make the most useful features? Welcome to feature engineering where we’ll discuss good vs. bad features and how you can pre-process and change them for optimal use in your models.

  • Introduction to Feature Engineering – (Video) 52 sec
PLATFORM:COURSERA
PROVIDER:GOOGLE CLOUD
COST:FREE (AUDIT)
DIFFICULTY:INTERMEDIATE
LANGUAGE:ENGLISH
CERTIFICATE:AVAILABLE (PAID)
EFFORT:11 HOURS
START DATE:ON GOING
DURATION:1 WEEK
HURRY UP TO ENROLL FOR FREE.

Raw Data to Features

Feature engineering is often the longest and most difficult phase during building your Machine Learning projects. In the facility engineering process, you start with your raw data and use your own domain knowledge to create features that will work your machine learning algorithms. In this module we find out what a good feature is and how to represent them in our Machine Learning model.

  • Raw Data to Features – (Video) 2 min
  • Good vs Bad Features – (Video) 2 min
  • Quiz: Features are Related to the Objective – (Video) 3 min
  • Features Known at Prediction-time – (Video) 3 min
  • Quiz: Features are Knowable at Prediction Time – (Video) 4 min
  • Features should be Numeric – (Video) 27 sec
  • Quiz: Features Should be Numeric – (Video) 5 min
  • Features Should Have Enough Examples – (Video) 1 min
  • Quiz: Features Should Have Enough Examples (p1) – (Video) 2 min
  • Quiz: Features Should Have Enough Examples (p2) – (Video) 2 min
  • Bringing Human Insight – (Video) 27 sec
  • Representing Features – (Video) 8 min
  • ML vs Statistics – (Video) 3 min
  • Lab Solution: Improve model accuracy with new features – (Video) 12 min
  • Raw Data to Features – (Practice Exercise) 6 min
  • Representing Features – (Practice Exercise) 6 min

Pre-processing and Feature Creation

This part of the Feature Creation module covers pre-processing and feature creations which are data processing techniques and can help you design a set feature for a machine learning system.

  • Preprocessing and Feature Creation – (Video) 6 min
  • Beam and Dataflow – (Video) 9 min
  • Lab Intro: Simple Dataflow Pipeline – (Video) 19 sec
  • Lab Solution: Simple Dataflow Pipeline – (Video) 6 min
  • Data Pipelines that Scale – (Video) 5 min
  • Lab Intro: MapReduce in Dataflow – (Video) 33 sec
  • Lab Solution: MapReduce in Dataflow – (Video) 3 min
  • Preprocessing with Cloud Dataprep – (Video) 6 min
  • Lab Intro: Computing Time-Windowed Features in Cloud Dataprep – (Video) 10 min
  • Lab Solution: Computing Time-Windowed Features in Cloud Dataprep – 36 sec
  • Preprocessing and Feature Creation – (Video) 14 min
  • Apache Beam and Cloud Dataflow – (Video) 14 min
  • Preprocessing with Cloud Dataprep – (Video) 4 min

Feature Crosses

In traditional machine learning, feature crosses did not play significant a role, but in modern times Machine Learning methods, feature crosses are an invaluable part of our toolkit. In this Feature Engineering module, you will learn the methods to identify the problems where feature crosses can be a powerful tool to help machines learn.

  • Introducing Feature Crosses – (Video) 58 sec
  • What is a Feature Cross? – (Video) 5 min
  • Discretization – (Video) 1 min
  • Memorization vs. Generalization – (Video) 4 min
  • Taxi colors – (Video) 4 min
  • Lab Intro: Feature Crosses to create a good classifier – (Video) 26 sec
  • Lab Solution: Feature Crosses to create a good classifier – (Video) 6 min
  • Sparsity + Quiz – (Video) 5 min
  • Lab Intro: Too Much of a Good Thing – (Video) 31 sec
  • Lab Solution: Too Much of a Good Thing – (Video) 7 min
  • Implementing Feature Crosses – (Video) 5 min
  • Embedding Feature Crosses – (Video) 9 min
  • Where to Do Feature Engineering – (Video) 6 min
  • Feature Creation in TensorFlow – (Video) 2 min
  • Feature Creation in DataFlow – (Video) 2 min
  • Lab Intro: Improve ML Model with Feature Engineering – (Video) 42 sec
  • Lab Solution (p1): ML Fairness Debrief – (Video) 3 min
  • Lab Solution (p2): Improve ML Model with Feature Engineering – (Video) 20 min
  • Feature crosses – (Practice Exercise) 8 min

TF Transform

TensorFlow Transform (tf.Transform) is a library used to pre-process the data with TensorFlow. tf.Transform is useful in pre-processing data that requires a full pass of the data, such as: – normalizing an input value by mean and standard deviation – integerizing the vocabulary by looking all the input examples for values – bucketizing inputs based on the observed data distribution. In this module you will get to explore the use cases for tf.Transform.

  • Introducing TensorFlow Transform – (Video) 25 sec
  • TensorFlow Transform – (Video) 8 min
  • Analyze phase – (Video) 3 min
  • Transform phase – (Video) 4 min
  • Supporting serving – (Video) 3 min
  • Lab Intro: Exploring tf.transform – (Video) 1 min
  • Lab Solution: Exploring tf.transform – (Video) 19 min
  • tf.transform – (Practice Exercise) 6 min

Summary of Feature Engineering

Here revision of the major points and topics that you have learned in each module above on Feature Engineering is done: Selecting Good Features, Pre-processing at Scale, Using Feature Crosses, and Practicing with TensorFlow.

  • Summary – (Video) 3 min

LEAVE A REPLY

Please enter your comment!
Please enter your name here