Machine Learning for SEOs - SMXL
•Download as PPTX, PDF•
1 like•262 views
The document discusses machine learning and how it can be used by SEOs. It defines machine learning and provides examples of applications like spam filtering, product recommendations, and home price predictions. The document encourages readers to think of problems machine learning could solve using available data and models. Specific opportunities for SEOs are discussed, like predicting customer churn, title tag optimization, and log file analysis. Readers are provided resources for learning machine learning.
Report
Share
Machine Learning for SEOs - SMXL
- 1. Machine Learning for SEOs @BritneyMuller Senior SEO Scientist
- 2. To do better, we must think differently
- 5. Machine learning can be your laser beam!
- 6. You don’t have to be a Data Scientist to think of the next brilliant ML application!
- 8. Let’s solve a problem together using ML How can we help solve a customer churn problem?
- 9. Problem: Customer Churn To do this, you need to know: 1. What Machine Learning is 2. How to identify ML opportunities 3. Specific examples of ML in action 4. How to integrate ML into your thinking process [Framework] Solution: What sort of data + model might you need to do this?
- 10. What is Machine Learning? Machine Learning is a subset of AI that combines statistics & programming to give computers the ability to “learn” without explicitly being programmed.
- 12. If Machine Learning was a car data would be the fuel.
- 14. For example, teachers, nurses, childcare ML doesn’t solve well for soft/people skills:
- 15. Safe & effective ML requires diversity
- 19. Problem: Customer Churn To do this, you need to know: 1. What Machine Learning is 2. How to identify ML opportunities 3. Specific examples of ML in action 4. How to integrate ML into your thinking process [Framework] Solution: What sort of data + model might you need to do this?
- 20. Classification: Spam / Not Spam Association: If milk is in someone’s cart, they are 80% more likely to buy bread. Regression: Prediction – Home price based on the number of bedrooms, bathrooms, m2 ,etc.
- 26. #TTTLIVE19
- 27. #TTTLIVE19
- 30. Problem: Customer Churn To do this, you need to know: 1. What Machine Learning is 2. How to identify ML opportunities 3. Specific examples of ML in action 4. How to integrate ML into your thinking process [Framework] Solution: What sort of data + model might you need to do this?
- 31. Machine Learning will free us up to do more strategic work.
- 32. Source: news.efinancialcareers.com/ca-en/285249/machine-learning-and-big-data-j-p-morgan “Machines have the ability to quickly analyze news feeds and tweets, process earnings statements, scrape websites, and trade on these instantaneously.”
- 33. Video Generation
- 38. Machine Learning is becoming more accessible & will free us up to work on higher level strategy.
- 41. Machine Learning will free us up to do more strategic work.
- 47. Machine Learning is becoming more accessible & will free us up to work on higher level strategy.
- 48. @BritneyMuller
- 49. We have only scratched the surface title tag optimization deduping questions (Quora, Stack Overflow) log file analysis parsing text into entities (ex. insurance forms) traffic predictions deeper user engagement insights website audit insights automatic website fixes instant alerts on website errors + SERP flux
- 50. Problem: Customer Churn To do this, you need to know: • What Machine Learning is • How to identify when you can use ML to solve problems • Specific examples of ML in action • How to integrate ML into your thinking process [Framework] Solution: What sort of data + model might you need to do this?
- 51. Problem: Customer Churn To do this, you need to know: 1. What Machine Learning is 2. How to identify ML opportunities 3. Specific examples of ML in action 4. How to integrate ML into your thinking process [Framework] Solution: What sort of data + model might you need to do this?
- 53. 1. What would you like to solve for? 2. Do you have labeled data to help train a model? 3. If not, can you start to collect data to help solve for your problem? 4. If not, consider what data you currently have and what you could solve with it. Simple ML Framework
- 55. Let’s solve a problem together using ML How can we help solve a customer churn problem?
- 56. Let’s solve a problem together using ML What would we want a model to do to prevent churn?
- 57. Classification: Spam / Not Spam Association: If milk is in someone’s cart, they are 80% more likely to buy bread. Regression: Prediction – Home price based on the number of bedrooms, bathrooms, m2 ,etc.
- 58. Let’s solve a problem together using ML What kind of data would we need to train a model to do that?
- 64. Download GSC data Get low CTR pages Scrape page titles Find top keywords per page Find pages missing top keywords in their title Rewrite/add keyword to the title
- 67. Getting Started • Search ‘Harvard CS109’ in GitHub • Learn Python in 10 Mins • Google CodeLabs – Break things!!! • MNist --The “Hello World!” of Machine Learning • Colab Notebooks OR Jupyter Notebooks • Learn With Google AI • Image-net.org • Kaggle • MonkeyLearn
- 68. Top ML Books
- 69. Free ML Books: bit.ly/free-ml-books • Statistics: New Foundations, Toolbox, and Machine Learning Recipes • Classification and Regression in a Weekend • Online Encyclopedia of Statistical Science • Azure Machine Learning in a Weekend • Enterprise AI - An Application Perspective • Applied Stochastic Processes (With a free Data Science Central account)
- 71. • Yearning Learning (free book preview by Andre Ng) • Neural Networks & Deep Learning • Correlation vs Causation (by Dr. Pete!) • Exploring Word2Vec • The Zipf Mystery • BigML • Targeting Broad Queries in Search • Project Mosaic Books • Algorithmia • How to eliminate bias in data driven marketing • TensorFlow Dev Summit 2018 [videos] • NLP Sentiment Analysis • Talk 2 Books • The Shallowness of Google Translate • TF-IDF • LSI • LDA • Learn Python • Massive Open Online Courses • Coursera Machine Learning • RAY by Professors at UC Berkeley Advanced Resources
- 72. People to follow
- 73. ML for SEOs Takeaways: • ML is programming + statistics that gives computers the ability to learn • An ML model is only as good as its training data • ML opportunities occur where available data can be used to predict, classify, discover associations/insights, etc. • Consider the data you have & what you could do with it • Diversity is paramount in ML • YOU can create an ML model today!!!
- 74. The Data Science Team at Moz is innovating in this space to make your journey from data to insights more efficient
- 76. Thank You!
- 78. Wut?
- 79. Sentiment Analysis Named entity recognition Question and answering Classification Machine translation Summarization Sentence disambiguation
- 80. BERT combines and outperforms 10+ of the common NLP tools
- 81. BERT combines and outperforms 10+ of the common NLP tools A pre-trained BERT model can be finetuned with just one additional output layer to create a SOTA model for wide range tasks such as question answering. Sound familiar??
- 83. + + + +
- 85. What BERT can’t do
- 86. You can play around with BERT today: