Regularization – definition and python example.
In Machine Learning, very often the task is to fit a model to a set of training data and use the fitted model to make predictions or classify new (out of sample) data points. Sometimes model fits the training data very well but does not well in predicting out of sample data points. A model may be too complex and overfit or too simple and underfit, either way giving poor predictions.
What is Regularization?
Regularization is a way to avoid overfitting by penalizing high regression coefficients, it can be seen as a way to control the tradeoff between bias and variance in favor of an increased generalization. In simple terms, it reduces parameters and simplifies the model or selects the preferred level of model complexity so it is better at predictinggeneralizing.
To apply regularization two things are required:
 A way of quantifying how a good model is eg. crossvalidation
 A tuning parameter which enables changing the complexity of the model
How does Regularization work?
In order to find the best model, the common method in machine learning is to define a loss function that describes how well the model fits the data. The ultimate goal is to minimize this loss function. Regularization is the process of adding a tuning parameter to a model, this is most often done by adding a constant multiple to an existing weight vector. The model predictions should then minimize the mean of the loss function calculated on the regularized training set.
Most often used regularization methods:

Ridge Regression(L2)

Lasso (L1) – “Least Absolute Selection and Shrinkage Operator”

ElasticNet
Example code of L1 regularization using Python:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 
from sklearn.linear_model import LogisticRegression from sklearn import datasets from sklearn.cross_validation import train_test_split import numpy as np data = datasets.load_iris() X = data['data'] y = data['target'] X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.4, random_state=0) for Coef in np.arange(0.1, 1,0.1): clf = LogisticRegression(penalty='l1', C=Coef) clf.fit(X_train, y_train) print('C:', Coef) print('Accuracy:', clf.score(X_test, y_test)) print('') 
Was the above useful? Please share with others on social media.
If you want to look for more information, check some free online courses available at coursera.org, edx.org or udemy.com.
Recommended reading list:
Data Science from Scratch: First Principles with Python Data science libraries, frameworks, modules, and toolkits are great for doing data science, but they’re also a good way to dive into the discipline without actually understanding data science. In this book, you’ll learn how many of the most fundamental data science tools and algorithms work by implementing them from scratch. If you have an aptitude for mathematics and some programming skills, author Joel Grus will help you get comfortable with the math and statistics at the core of data science, and with hacking skills you need to get started as a data scientist. Today’s messy glut of data holds answers to questions no one’s even thought to ask. This book provides you with the knowhow to dig those answers out. Get a crash course in Python Learn the basics of linear algebra, statistics, and probability—and understand how and when they're used in data science Collect, explore, clean, munge, and manipulate data Dive into the fundamentals of machine learning Implement models such as knearest Neighbors, Naive Bayes, linear and logistic regression, decision trees, neural networks, and clustering Explore recommender systems, natural language processing, network analysis, MapReduce, and databases 

Practical Statistics for Data Scientists: 50 Essential Concepts Statistical methods are a key part of of data science, yet very few data scientists have any formal statistics training. Courses and books on basic statistics rarely cover the topic from a data science perspective. This practical guide explains how to apply various statistical methods to data science, tells you how to avoid their misuse, and gives you advice on what's important and what's not. Many data science resources incorporate statistical methods but lack a deeper statistical perspective. If you’re familiar with the R programming language, and have some exposure to statistics, this quick reference bridges the gap in an accessible, readable format. With this book, you’ll learn: Why exploratory data analysis is a key preliminary step in data science How random sampling can reduce bias and yield a higher quality dataset, even with big data How the principles of experimental design yield definitive answers to questions How to use regression to estimate outcomes and detect anomalies Key classification techniques for predicting which categories a record belongs to Statistical machine learning methods that “learn” from data Unsupervised learning methods for extracting meaning from unlabeled data 

Doing Data Science: Straight Talk from the Frontline Now that people are aware that data can make the difference in an election or a business model, data science as an occupation is gaining ground. But how can you get started working in a wideranging, interdisciplinary field that’s so clouded in hype? This insightful book, based on Columbia University’s Introduction to Data Science class, tells you what you need to know. In many of these chapterlong lectures, data scientists from companies such as Google, Microsoft, and eBay share new algorithms, methods, and models by presenting case studies and the code they use. If you’re familiar with linear algebra, probability, and statistics, and have programming experience, this book is an ideal introduction to data science. Topics include: Statistical inference, exploratory data analysis, and the data science process Algorithms Spam filters, Naive Bayes, and data wrangling Logistic regression Financial modeling Recommendation engines and causality Data visualization Social networks and data journalism Data engineering, MapReduce, Pregel, and Hadoop 

The Data Science Handbook: Advice and Insights from 25 Amazing Data Scientists The Data Science Handbook contains interviews with 25 of the world s best data scientists. We sat down with them, had indepth conversations about their careers, personal stories, perspectives on data science and life advice. In The Data Science Handbook, you will find war stories from DJ Patil, US Chief Data Officer and one of the founders of the field. You ll learn industry veterans such as Kevin Novak and Riley Newman, who head the data science teams at Uber and Airbnb respectively. You ll also read about rising data scientists such as Clare Corthell, who crafted her own open source data science masters program. This book is perfect for aspiring or current data scientists to learn from the best. It s a reference book packed full of strategies, suggestions and recipes to launch and grow your own data science career. 

Introduction to Machine Learning with Python: A Guide for Data Scientists Machine learning has become an integral part of many commercial applications and research projects, but this field is not exclusive to large companies with extensive research teams. If you use Python, even as a beginner, this book will teach you practical ways to build your own machine learning solutions. With all the data available today, machine learning applications are limited only by your imagination. You’ll learn the steps necessary to create a successful machinelearning application with Python and the scikitlearn library. Authors Andreas Müller and Sarah Guido focus on the practical aspects of using machine learning algorithms, rather than the math behind them. Familiarity with the NumPy and matplotlib libraries will help you get even more from this book. With this book, you’ll learn: Fundamental concepts and applications of machine learning Advantages and shortcomings of widely used machine learning algorithms How to represent data processed by machine learning, including which data aspects to focus on Advanced methods for model evaluation and parameter tuning The concept of pipelines for chaining models and encapsulating your workflow Methods for working with text data, including textspecific processing techniques Suggestions for improving your machine learning and data science skills 