SearchLove San Diego 2017 | Michael King | Machine Doing

132
Managing Director 347-391-4266 [email protected] MACHINE DOING 12/01/2016 Michael King

Transcript of SearchLove San Diego 2017 | Michael King | Machine Doing

Managing Director

347-391-4266

[email protected]

MACHINE DOING

12/01/2016

Michael King

IPULLRANK.COM @ IPULLRANK

Agenda

Machine Learning Doomsday

ML vs DL vs AI?

Marketing Use Cases

Models & Use Cases

Tools For Marketers

Wrapping Up

Real World Examples

Machine Learning Doomsday

Subtitle goes here.

IPULLRANK.COM @ IPULLRANK

Smart People are Scared of Artificial Intelligence

IPULLRANK.COM @ IPULLRANK

Really Smart People…

It’s All Olivia Pope’s dad’s fault

Either the Robots Enslave Us

Or Kill Us All

Or Evolve to a Point We Can’t Understand

Or we achieve singularity

IPULLRANK.COM @ IPULLRANK

Singularity is Considered a Very Real Theory

Ray Kurzweil believes that we will achieve singularity by 2045.

No matter what, it’s all Olivia Pope’s dad’s fault

IPULLRANK.COM @ IPULLRANK

No Matter What Larry & Sergey are All Set Though

IPULLRANK.COM @ IPULLRANK

Don’t Forget Isaac Asimov

IPULLRANK.COM @ IPULLRANK

Machine Learning Can Write Copy For you

There is a sub-field of artificial intelligence called Natural Language Generation that has made the concept of content spinning a lot more viable and has been used for sports recaps and financial reports.

IPULLRANK.COM @ IPULLRANK

But It Can Also Fuck It Up

IPULLRANK.COM @ IPULLRANK

AI Is Gonna Steal Your Job?

One of the more common fears of middle America around the idea of artificial intelligence is that robots will replace humans in their jobs.

IPULLRANK.COM @ IPULLRANK

Obama Had Some Measured Thoughts On His Way Out

The real fear of machine learning and artificial intelligence should be its ability to reflect and amplify our biases and the lack of diversity of the people creating it.

IPULLRANK.COM @ IPULLRANK

In the meantime though, it can get you a date

(h/t @goutaste)

IPULLRANK.COM @ IPULLRANK

And you can go there in a self-driving Uber Lyft

Machine Learning vs. Deep Learning vs. AI

The Core Concepts

IPULLRANK.COM @ IPULLRANK

They Are Not the Same Thing

IPULLRANK.COM @ IPULLRANK

AI is Comprised of Many Disciplines

Deep Learning is a subset of Machine Learning is a subset of Artificial Intelligence.

AI many branches of which machine learning is a core branch that we can execute.

Artificial Intelligence as it is represented in sci-fi is “general” artificial intelligence. What we have achieved so far is “narrow” artificial intelligence.

IPULLRANK.COM @ IPULLRANK

Types of Artificial Intelligence Explained Using “The Lawnmower Man”

Narrow Artificial IntelligenceMachines that can do a specific task or series of tasks exceedingly well and very efficiently.

General Artificial IntelligenceA machine that is as smart as a human in that it can take in new situations and make decisions.

Artificial SuperintelligenceA machine that is potentially orders of magnitude smarter than a human in all categories simultaneously

IPULLRANK.COM @ IPULLRANK

Experts Disagree on When General Intelligence Will Happen

The primary i ssue keeping this from happening is computing power.

IPULLRANK.COM @ IPULLRANK

Experts Disagree on When General Intelligence Will Happen

The primary i ssue keeping this from happening is computing power.

IPULLRANK.COM @ IPULLRANK

Accelerating Moore’s Law

Google has been working on quantum computing to accelerate Moore’s Law

IPULLRANK.COM @ IPULLRANK

100mmtimes faster than a classical computer by

using a D-Wave quantum computer

NewScientist.com

Ok. So, What Is Machine Learning?

“Machine learning is a type of artificial intelligence that provides computers with the ability to learn without being explicitly programmed.”

IPULLRANK.COM @ IPULLRANK

Supervised Learning

The machine looks for patterns that match the labeled data that you provide and classifies new data based on that.

IPULLRANK.COM @ IPULLRANK

Unsupervised Learning

The machine identifies patterns in the data and creates clusters based on what it finds.

IPULLRANK.COM @ IPULLRANK

Reinforcement Learning

With reinforcement learning, the model is continually trained based on new data thereby improving the classifier’s ability to perform.

And Deep Learning?

“Deep Learning is a subfield of machine learning concerned with algorithms inspired by the structure and function of the brain called artificial neural networks.”

That’s not what we’re talking about today.

Machine Learning vs. Statistics

Machine Learning learns from data without relying on rules-based programming, statistical modeling identifies relationships in the form of mathematical equations.

IPULLRANK.COM @ IPULLRANK

All Values vs. Linear Representation

Machine Learning examines all potential values based on probability whereas statistics looks for a linear function to

describe the trend.

IPULLRANK.COM @ IPULLRANK

Machine Learning is the “Growth Hacking” of the Statistics World

However, in some ways machine learning and statistics are so similar that many statisticians just feel as though machine learning is just a rebranding of what they do much like “growth hacking” is just a rebranding of marketing.

IPULLRANK.COM @ IPULLRANK

The Machine Learning Process

GET & PREPARE YOUR DATA

You identify and clean your dataset in preparation for

solving the machine learning problem

CHOOSE YOUR MODEL TRAIN YOUR CLASSIFIER

You chose the algorithm or model that you believe will

yield the best results then run it in order to train your

classifier.

SCORE AND EVALUATE

You score the accuracy and precision of the classifier and

test it against other algorithms to see what

performs best.

PREDICT OR IDENTIFY OUTCOMES

Once you are happy with the results, you use the classifier

moving forward to make conclusions about new data.

IPULLRANK.COM @ IPULLRANK

Car Rental Example

This is an example of how you could predict the

demand of cars for a car rental company. It follows

the same framework.

Marketing Use Cases

Subtitle goes here.

IPULLRANK.COM @ IPULLRANK

Predictive Analytics

IPULLRANK.COM @ IPULLRANK

Marketing Campaign Performance Prediction

IPULLRANK.COM @ IPULLRANK

Customer Churn Prediction

IPULLRANK.COM @ IPULLRANK

Personalization

IPULLRANK.COM @ IPULLRANK

Customer Segmentation

IPULLRANK.COM @ IPULLRANK

Natural Language Processing

IPULLRANK.COM @ IPULLRANK

Clustering & Classifying Keywords

http://ipullrank.com/clustering-vs-classification-speed-keyword-research/

IPULLRANK.COM @ IPULLRANK

Clustering & Classifying Keywords

http://ipullrank.com/clustering-vs-classification-speed-keyword-research/

Follow Vicky Qian

@vickyqian24

IPULLRANK.COM @ IPULLRANK

Sentiment Analysis

IPULLRANK.COM @ IPULLRANK

Natural Language Generation

IPULLRANK.COM @ IPULLRANK

Computer Vision

There are services that leverage machine learning and computer vision to identify objects in pictures.

IPULLRANK.COM @ IPULLRANK

Chatbots

IPULLRANK.COM @ IPULLRANK

Training Chatbots

Tra ining chatbots is similar to training ML classifiers in that you take a knowledge base and run it through NLP then tune i t with regard to conversations.

Real World Examples

Some things we work on

IPULLRANK.COM @ IPULLRANK

We Re-Ranked the Inc. 500

IPULLRANK.COM @ IPULLRANK

Follow Up Blog Content

IPULLRANK.COM @ IPULLRANK

Company-level Report

IPULLRANK.COM @ IPULLRANK

Retargeting Ads

IPULLRANK.COM @ IPULLRANK

Super-specific Retargeting Ads

IPULLRANK.COM @ IPULLRANK

We Built a Simple Marketing Automation System

IPULLRANK.COM @ IPULLRANK

Each Contact Has a Unique URL

IPULLRANK.COM @ IPULLRANK

Integrates with Reply

IPULLRANK.COM @ IPULLRANK

LinkedIn Sales Navigator

IPULLRANK.COM @ IPULLRANK

Prospectify for finding Emails Quickly

IPULLRANK.COM @ IPULLRANK

Salesperson Writes Mail Merge Templates

IPULLRANK.COM @ IPULLRANK

42% Open Rate!

732 Leads

IPULLRANK.COM @ IPULLRANK

The Methodology is the Machine Learning Part

We took all available domain-level link features for the Searchmetrics losers and winners and figured out (5-fold cross validation, random forest and lasso) which ones correlated best with the results and then used that model to re-rank the Inc. 500. (I probably shoulda asked Marcus for more data, but whatever).

IPULLRANK.COM @ IPULLRANK

Methodology behind the Vector Report

We broke it into two types of machine learning questions. Classification and Logistic Regression to predict the probability of continued visibility in Organic Search.

Goal: identify SEO winners and losers and

predict a site’s performance in SEO

Classification

Random Forest

Gradient Boosting Machine

Support Vector Machine

Logistic Regression Regularization

IPULLRANK.COM @ IPULLRANK

Have You Met @tomcritchlowbot?

IPULLRANK.COM @ IPULLRANK

The Bot wit the Solid Delivery.

IPULLRANK.COM @ IPULLRANK

This is Twitter bot built from Markov Chains

IPULLRANK.COM @ IPULLRANK

Adwords Scripts

http://searchengineland.com/machine-learning-adwords-scripts-google-prediction-api-217936

IPULLRANK.COM @ IPULLRANK

Programmatic Display

Lookalike moderlign

Models

Types of Models when you should Use Them

IPULLRANK.COM @ IPULLRANK

There are Tons of Different Models

Your best bet is to test and learn.

IPULLRANK.COM @ IPULLRANK

Seriously Tonnnnnns

IPULLRANK.COM @ IPULLRANK

The Uses of Each Type Are Difficult to Memorize

IPULLRANK.COM @ IPULLRANK

Models & Use Cases

Random ForestLead Qualification

Logistic RegressionCustomer ChurnPrediction

Decision TreesCustomer ChurnPrediction

IPULLRANK.COM @ IPULLRANK

Models & Use Cases (Cont’d)

Support Vector MachinesTextCategorization

AprioriMarket Basket Analysis(Amazon)

Naïve BayesSentiment AnalysisRecommendation SystemsSpam Classification

IPULLRANK.COM @ IPULLRANK

K-Fold Cross Validation

Try out a model and validate it using k-fold cross validation.

IPULLRANK.COM @ IPULLRANK

How to Choose a Machine Learning Model

https://docs.microsoft.com/en-us/azure/machine-learning/machine-learning-algorithm-choice

Tools for Marketers

Subtitle goes here.

IPULLRANK.COM @ IPULLRANK

IPULLRANK.COM @ IPULLRANK

IPULLRANK.COM @ IPULLRANK

yHat Science Ops

Open source machine learning and data visualization for novice and expert.

Most machine learning is done in R or Python, but those are programming languages.

IPULLRANK.COM @ IPULLRANK

yHat Science Ops

yHat allows you to deploy machine learning models as REST APIs that can then be integrated with your site like any other API.

IPULLRANK.COM @ IPULLRANK

Beeswax Bidder-as-a-Service

Beeswax allows you to set up custom models to run your Display RTB campaigns.

Those are tools that allow marketers to take control with a data scientist.

IPULLRANK.COM @ IPULLRANK

mTurk - Labeling Data for Supervised Learning

Exploratory Data Analysis helps identifying general patterns in the data and serve as initial explorations of correlations.

IPULLRANK.COM @ IPULLRANK

API.AI Generating Chatbots

Exploratory Data Analysis helps identifying general patterns in the data and serve as initial explorations of correlations.

IPULLRANK.COM @ IPULLRANK

NanoRep

Exploratory Data Analysis helps identifying general patterns in the data and serve as initial explorations of correlations.

IPULLRANK.COM @ IPULLRANK

MonkeyLearn & Orange

We wi ll primarily ta lk about MonkeyLearn and Orange as two tools marketers can use to do machine learning right now.

IPULLRANK.COM @ IPULLRANK

These Examples Use the Iris Petals Dataset

https://archive.ics.uci.edu/ml/datasets/Iris

IPULLRANK.COM @ IPULLRANK

Exploratory Data Analysis

Exploratory Data Analysis helps identifying general patterns in the data and serve as initial explorations of correlations.

IPULLRANK.COM @ IPULLRANK

Exploratory Data Analysis: Scatter Plot

Two-dimensional scatter plot shows class density.

IPULLRANK.COM @ IPULLRANK

Exploratory Data Analysis: Distributions

Compare the distributions of different type of i ris.

IPULLRANK.COM @ IPULLRANK

Classification Tree

Observe the pattern across nodes to discover important variables.

IPULLRANK.COM @ IPULLRANK

Predictive Text Classification

IPULLRANK.COM @ IPULLRANK

Import Text Mining Add-on

Install the free text mining add-on in order to use

Orange’s text mining capabilities.

IPULLRANK.COM @ IPULLRANK

Load and Preprocess Dataset

Preprocess text to find meaningful words only.

IPULLRANK.COM @ IPULLRANK

Word Cloud

Using the word cloud, we can determine the frequency of keywords in the list.

IPULLRANK.COM @ IPULLRANK

Hierarchical Clustering

We can use this to determine similarity in the corpus or dataset.

IPULLRANK.COM @ IPULLRANK

Hierarchical Clustering

Once we understand the hierarchy, we can dig into the documents in the viewer to see how the model has organized them.

IPULLRANK.COM @ IPULLRANK

Classification

IPULLRANK.COM @ IPULLRANK

SVM: Linear vs. Non-linear

Linear SVM often outperforms non-linear in text classification.

IPULLRANK.COM @ IPULLRANK

Confusion Matrix: Non-linear SVM

Send misclassified samples to corpus viewer.

IPULLRANK.COM @ IPULLRANK

Nearest Neighbors

IPULLRANK.COM @ IPULLRANK

Logistic Regression: Ridge vs. Lasso

Logistic regression with l2 penalty achieve higher accuracy.

IPULLRANK.COM @ IPULLRANK

Confusion Matrix: Lasso

Send misclassified samples to corpus viewer.

IPULLRANK.COM @ IPULLRANK

Naive Bayes

IPULLRANK.COM @ IPULLRANK

Compare Models

Linear SVM and logistic regression outperform the other two models.

IPULLRANK.COM @ IPULLRANK

Prediction

Predict with winning classifiers.

IPULLRANK.COM @ IPULLRANK

Prediction

SVM and logistic regression all hit 100% accuracy.

IPULLRANK.COM @ IPULLRANKMonkey Learn is a text mining cloud platform.

IPULLRANK.COM @ IPULLRANKMonkey Learn is a text mining cloud platform.

MonkeyLearn Now Works with Google Sheets

IPULLRANK.COM @ IPULLRANK

Monkey Learn: Train Category Tree

IPULLRANK.COM @ IPULLRANK

Monkey Learn: Tree Parameters

IPULLRANK.COM @ IPULLRANK

Monkey Learn: Classify with Category Tree

IPULLRANK.COM @ IPULLRANK

Codementor

IPULLRANK.COM @ IPULLRANK

Experfy

IPULLRANK.COM @ IPULLRANK

Kaggle

Wrapping Up

Who am I and where am I from?

IPULLRANK.COM @ IPULLRANK

I’M #ZORASDAD

First and foremost.

IPULLRANK.COM @ IPULLRANK

MY NAME IS MIKE KING

Razorfish, Publicis Modem alum

Full Stack Developer

Full Stack Marketer

Moz Associate

IPULLRANK.COM @ IPULLRANK

I Run a Better Marketing Agency Called iPullRank

IPULLRANK.COM @ IPULLRANK

We Do These Things

Content

StrategySEO Paid Media Machine

Learning

Marketing

Automation

Measurement

& Optimization

THAT’S ALL I’VE GOT

IPULLRANKhttp://ipullrank.com

THANK YOU

Michael King

Managing Director

(347) [email protected]

02/21/2017