Book Image

Machine Learning Quick Reference

By : Rahul Kumar
Book Image

Machine Learning Quick Reference

By: Rahul Kumar

Overview of this book

Machine learning makes it possible to learn about the unknowns and gain hidden insights into your datasets by mastering many tools and techniques. This book guides you to do just that in a very compact manner. After giving a quick overview of what machine learning is all about, Machine Learning Quick Reference jumps right into its core algorithms and demonstrates how they can be applied to real-world scenarios. From model evaluation to optimizing their performance, this book will introduce you to the best practices in machine learning. Furthermore, you will also look at the more advanced aspects such as training neural networks and work with different kinds of data, such as text, time-series, and sequential data. Advanced methods and techniques such as causal inference, deep Gaussian processes, and more are also covered. By the end of this book, you will be able to train fast, accurate machine learning models at your fingertips, which you can easily use as a point of reference.

Related Content you might be interested in

Current Title:

Small book image
Machine Learning Quick Reference
Rahul Kumar
Jan, 2019 | 294 pages
Small book image
Practical Time Series Analysis
Avishek Pal | PKS Prakash
Sep, 2017 | 244 pages
Small book image
Ensemble Machine Learning Cookbook
Vijayalakshmi Natarajan | Dipayan Sarkar
Jan, 2019 | 336 pages
Small book image
Hands-On Python for Finance
Krish Naik
Mar, 2019 | 378 pages
Table of Contents (18 chapters)
Title Page
Title Page
Copyright and Credits
Copyright and Credits
About Packt
About Packt
Contributors
Contributors
Preface
Preface
Free Chapter
1
Quantifying Learning Algorithms
Quantifying Learning Algorithms
Statistical models
Learning curve
Curve fitting
Statistical modeling – the two cultures of Leo Breiman
Training data development data – test data
Bias-variance trade off
Regularization
Cross-validation and model selection
Model selection using cross-validation
0.632 rule in bootstrapping
Model evaluation
Receiver operating characteristic curve
H-measure
Dimensionality reduction
Summary
2
Evaluating Kernel Learning
Evaluating Kernel Learning
Introduction to vectors
Linear separability
Hyperplanes 
SVM
Kernel trick
Kernel types
SVM example and parameter optimization through grid search
Summary
3
Performance in Ensemble Learning
Performance in Ensemble Learning
What is ensemble learning?
Bagging
Decision tree
Random forest algorithm
Boosting
Summary
4
Training Neural Networks
Training Neural Networks
Neural networks
Network initialization
Overfitting
Prevention of overfitting in NNs
Vanishing gradient 
Recurrent neural networks
Summary
5
Time Series Analysis
Time Series Analysis
Introduction to time series analysis
White noise
Random walk
Autoregression
Autocorrelation
Stationarity
AR model
Moving average model
Autoregressive integrated moving average
Optimization of parameters
Anomaly detection
Summary
6
Natural Language Processing
Natural Language Processing
Text corpus
TF-IDF
Sentiment analysis
Topic modeling 
The Bayes theorem
Summary
7
Temporal and Sequential Pattern Discovery
Temporal and Sequential Pattern Discovery
Association rules
Apriori algorithm
Frequent pattern growth
Summary
8
Probabilistic Graphical Models
Probabilistic Graphical Models
Key concepts
Bayes rule
Bayes network
Summary
9
Selected Topics in Deep Learning
Selected Topics in Deep Learning
Deep neural networks
Backward propagation
Forward propagation equation
Backward propagation equation
Parameters and hyperparameters
Bias initialization
Generative adversarial networks
Hinton's Capsule network
Summary
10
Causal Inference
Causal Inference
Granger causality
F-test
Graphical causal models
Summary
11
Advanced Methods
Advanced Methods
Introduction
Kernel PCA
Independent component analysis
Compressed sensing
Self-organizing maps
Bayesian multiple imputation
Summary
Other Books You May Enjoy
Other Books You May Enjoy
Leave a review - let other readers know what you think
Index
Index

Training data development data – test data

This is one of the most important steps of building a model and it can lead to lots of debate regarding whether we really need all three sets (train, dev, and test), and if so, what should be the breakup of those datasets. Let's understand these concepts.

After we have sufficient data to start modelling, the first thing we need to do is partition the data into three segments, that is, Training Set, DevelopmentSet, and Test Set:

Let's examine the goal of having these three sets:

  1. Training Set: The training set is used to train the model. When we apply any algorithm, we are fitting the parameter in the training set. In the case of a neural network, finding out about the weights takes place.

Let's say in one scenario that we are trying to fit polynomials of various degrees:

    • f(x) = a+ bx → 1st degree polynomial
    • f(x) = a + bx + cx2 → 2nd degree polynomial
    • f(x) = a + bx + cx2 + dx3 → 3rd degree polynomial

After fitting the model, we calculate the training error for all the fitted models:

We cannot assess how good the model is based on the training error. If we do that, it will lead us to a biased model that might not be able to perform well on unseen data. To counter that, we need to head into the development set.

  1. Developmentset: This is also called the holdout set or validation set. The goal of this set is to tune the parameters that we have got from the training set. It is also part of an assessment of how well the model is performing. Based on its performance, we have to take steps to tune the parameters. For example, controlling the learning rate, minimizing the overfitting, and electing the best model of the lot all take place in the development set. Here, again, the development set error gets calculated and tuning of the model takes place after seeing which model is giving the least error. The model giving the least error at this stage still needs tuning to minimize overfitting. Once we are convinced about the best model, it is chosen and we head toward the test set.
  1. Test set: The test set is primarily used to assess the best selected model. At this stage, the accuracy of the model is calculated, and if the model's accuracy is not too deviated from the training accuracy and development accuracy, we send this model for deployment.

Size of the training, development, and test set

Typically, machine learning practitioners choose the size of the three sets in the ratio of 60:20:20 or 70:15:15. However, there is no hard and fast rule that states that the development and test sets should be of equal size. The following diagram shows the different sizes of the training, development, and test sets:

Another example of the three different sets is as follows:

But what about the scenarios where we have big data to deal with? For example, if we have 10,000,000 records or observations, how would we partition the data? In such a scenario, ML practitioners take most of the data for the training set—as much as 98-99%—and the rest gets divided up for the development and test sets. This is done so that the practitioner can take different kinds of scenarios into account. So, even if we have 1% of data for development and the same for the test test, we will end up with 100,000 records each, and that is a good number.

Previous Section
End of Section 6
Next Section