Advanced Predictive Techniques with Scikit-Learn& TensorFlow

Improve the performance predictive models, build more complex models and use techniques to improve quality of your predi

Created byPackt Publishing

Last updated 12/2017

English

What you'll learn

Use ensemble algorithms to combine many individual predictors to produce better predictions.
Apply advanced techniques such as dimensionality reduction to combine features and build better models.
Evaluate models and choose the optimal hyper-parameters using cross-validation.
Learn the foundations for working and building models using Neural Networks.
Learn different techniques to solve problems that arise when doing Predictive Analytics in the real world

Course content

5 sections • 18 lectures • 3h 44m total length

The Course Overview4:38
This video provides an overview of the entire course.
How Ensemble Methods Work?9:21
Explain the general idea behind ensemble methods and discuss at a high level the intuition of the main ensemble methods - bagging, random forest and boosting.
Bagging, Random Forests, and Boosting for Regression11:58
Present with a practical example the procedure to build ensemble methods for regression tasks and compare the results of ensemble methods with other simpler methods.
Bagging, Random Forests, and Boosting for Classification14:51
Present with a practical example the procedure to build ensemble methods for classification tasks and compare the results of ensemble methods with other simpler methods.
Test your Knowledge

K-fold Cross-Validation11:26
Explain the main problem with hold out cross validation and explain the approach of K-fold cross-validation to solve this problem. Present how to do K-fold cross validation in scikit-learn
Comparing Models with K-fold Cross-Validation6:45
Show how k-fold cross-validation can be used to get a better assessment of the performance of the models and hence to make better model comparison.
Hyper-Parameter Tuning in scikit-learn13:00
Explain the need for hyper-parameter tuning when building predictive analytics models and show how we can combine k-fold cross-validation with grid search to do it.
Test your Knowledge

Feature Selection Methods18:16
Explain why we need different methods to distinguish between useful and useless features and explain the ones that will be used in the video: low variance, statistical tests, and RFE.
Dimensionality Reduction and PCA8:33
Explain the idea of dimensionality reduction and the intuition of the Principal Component Analysis method and show how to use this tool in scikit-learn.
Creating New Features11:21
Explain what is feature engineering and the different methods and approaches to create new features show with examples how to do it with the dataset used in the course.
Improving Models with Feature Engineering12:09
Show the practical use of the techniques shown in the section and show how to improve one of the models built in the previous sections using feature engineering.
Test your Knowledge

Introduction to Artificial Neural Networks15:11
Give an overview of Artificial Neural Networks and its basic components: the perceptron, show intuitively how to construct networks of perceptrons.
Elements of a Deep Neural Network Model19:02
Explain how to go from one single layer of perceptrons to a multi-layer model (deep model) then give a general overview of all the elements that need to be considered to train these types of models.
Installation and Introduction to TensorFlow9:02
Explain what is TensorFlow, what is used for and how to install the CPU version of the library
Core Concepts in TensorFlow20:40
Discuss the main concepts and objects used in TensorFlow, the goal is the viewer to have some familiarity with the way this library works.
Test your Knowledge

Predictions with TensorFlow - Introductory Example14:45
Present and introductory example to show how a simple predictive model looks like in TensorFlow.
Regression Using Deep Neural Networks11:36
Show how to build a regression model using a Deep Neural Network and a real-world dataset, show the necessary steps for reading a dataset, train and evaluate the model.
Classification with Deep Neural Networks11:56
Show how to build a classification model using a deep neural network and a real-world dataset, show the necessary steps for reading a dataset, train and evaluate the model.
Test your Knowledge

Requirements

This course presents some of the most advanced Predictive Analytics tools, models, and techniques currently having a big impact on every industry. The main goal is to show the viewer how to improve the performance of predictive models—firstly, by showing how to build more complex models and secondly, by showing how to use related techniques that dramatically improve the quality of predictive models.

Description

Ensemble methods offer a powerful way to improve prediction accuracy by combining in a clever way predictions from many individual predictors. In this course, you will learn how to use ensemble methods to improve accuracy in classification and regression problems.

When using Predictive Analytics to solve actual problems, besides models and algorithms there are many other practical considerations that must be considered like which features should I use, how many features are enough, should I create new features, how to combine features to give the same underlying information, which hyper-parameters should I use? We explore topics that will help you answer such questions.

Artificial Neural Networks are models loosely based on how neural networks work in a living being. These models have a long history in the Artificial Intelligence community with ups and downs in popularity. Nowadays, because of the increase in computational power, improved methods, and software enhancements, they are popular again and are the basis for advanced approaches such as Deep Learning. This course introduces the use of Deep Learning models for Predictive Analytics using the powerful TensorFlow library.

About the Author :

Alvaro Fuentes is a Data Scientist with an M.S. in Quantitative Economics and a M.S. in Applied Mathematics with more than 10 years of experience in analytical roles. He worked in the Central Bank of Guatemala as an Economic Analyst, building models for economic and financial data. He founded Quant Company to provide consulting and training services in Data Science topics and has been a consultant for many projects in fields such as; Business, Education, Psychology and Mass Media. He also has taught many (online and in-site) courses to students from around the world in topics like Data Science, Mathematics, Statistics, R programming and Python.

Alvaro Fuentes is a big Python fan and has been working with Python for about 4 years and uses it routinely for analyzing data and producing predictions. He also has used it in a couple of software projects. He is also a big R fan, and doesn't like the controversy between what is the “best” R or Python, he uses them both. He is also very interested in the Spark approach to Big Data, and likes the way it simplifies complicated things. He is not a software engineer or a developer but is generally interested in web technologies.

He also has technical skills in R programming, Spark, SQL (PostgreSQL), MS Excel, machine learning, statistical analysis, econometrics, mathematical modeling.

Predictive Analytics is a topic in which he has both professional and teaching experience. Having solved practical problems in his consulting practice using the Python tools for predictive analytics and the topics of predictive analytics are part of a more general course on Data Science with Python that he teaches online.

Who this course is for:

The course is for data analysts or data scientists, software engineers, and developers interested in learning advanced Predictive Analytics with Python. Business analysts/business Intelligence experts who would like to learn how to go from basic predictive models to building advanced models to produce better predictions will also find this course indispensable.
Knowledge of Python and familiarity with its Data Science Stack are assumed. Additionally, an understanding of the basic concepts of predictive analytics and how to use basic predictive models is also necessary to take full advantage of this course.

Advanced Predictive Techniques with Scikit-Learn& TensorFlow

What you'll learn

Explore related topics

Course content

Ensemble Methods for Regression and Classification4 lectures • 41min

Cross-Validation and Parameter Tuning3 lectures • 31min

Working with Features4 lectures • 50min

Introduction to Artificial Neural Networks and TensorFlow4 lectures • 1hr 4min

Predictive Analytics with TensorFlow and Deep Neural Networks3 lectures • 38min

Requirements

Description

Who this course is for: