It's real google will pay you to do what you love - monetize your blog and YouTube channel

Monetize your blog and YouTube channel

It's true it's real google will actually pay you to do what you love the most. I have been blogging about learning to code, japan, and Pokémon GO and I just got paid by google - my first ad revenue is real and in the bank. How did I do it?

• Write about what you love. Your niche may be small but the authenticity of your writing, opinions and details make a huge difference. My blog audience is small, like in the hundreds and thousands but because my posts are relevant users end up enjoying the ads they are served (udacity, amazon web services). Despite not being able to earn much from CPM (based on 1000s impression), my blog's click through rare is high. CPC is a valuable earning 
• Be detail oriented. I am no internet sensation, I am not viral. How do I deliver value to my readers? By being thorough and detailed in my presentation of observations and the follow up analyses.
• Be compliant be 100% compliant. When Google Adsense first came out, I was young and had an account that I played with. I tested my site and clicked on the ads and was blocked forever. I probably can contact them to unblock because it has been 10 years but the reality is grim Google is hard to reach and bans are permanent to deter opportunists. They really mean it. If you are not compliant, displayed fraudulent info, violated copy right content on YouTube, click on your own ads Google's algorithm will find you and terminate your entire account and stop monetization on all channels. Google's algorithm has proven to be intelligent (machine learning), complex (deep learning) and capable ( big data). You will be caught. 
• Follow a schedule. Smartly timed intervals of updates is desirable. No one wants to be overwhelmed by spam or visit your site or channel and find nothing new 3 times in a row
• Have a brand or a style. All successful viral influencers have a brand or a style that is extremely distinctive. Think Justin Bieber, he has a very specific hair cut and look, and he only sings certain cheesy songs. But that makes him extremely recognizable. Bread girl on Instagram has a repeatable viral machine : putting her face in different types of bread. While the safety and the usefulness of such an act is questionable, its success is very repeatable. It's a machine! We may not have that money making machine but we can think and design our brand message. For me, this blog is about Silicon Valley tech lifestyle so it has gadgets, Silicon Valley jobs, startup tax and logistics. I cannot post food recipes here unless it is about how to make a Star Wars cake. Clam chowder is irrelevant here and will make my readers question my blog. It's 
• Analyze what works. My Pokémon GO posts on blogger generate thousands of views but no one watch my YouTube Pokémon GO videos. So I have to post more Pokémon on the blog and focus more on learn to code tutorials on my YouTube channel. Find what works, optimize and then repeat success make it better

Machine Learning code pattern 01

Code pattern 01 numpy use array().T to get matrix transpose

Example:

X = [1,2,3]
XT = array(X).T

Machine Learning Concepts 01

 There are three main machine learning styles:

1.  Supervised learning 
2.  Unsupervised learning
3.  Reninforcement learning

Udacity Machine learning Nanodegree Syllabus and Summary

udacity nanodegree - becoming a machine learning engineer. This is my personal notes summarizing what I learned from the section, consider it my personal study notes. The part I labeled syllabus is the actual outline of the course (e.g. I try to use section title as the syllabus section title)

Section supervised learning
Sub section artificial neutral networks
Sub section neural networks
     how does a neuron work (illustrated)
     How artificial neural network works (illustrated) perceptron
     A group of inputs, each with a weights, processed by the network and output 1 of and only if a preconfigured threshold is met
     Artificial neural networks can be tuned

• Think of inputs as signals with different strengths, weights as sensitivity to those strengths. Can be tuned and adjusted to computer variety of tasks. Hence a collection of perceptroncomputing units is powerful
• The weights sum of all the inputs is called the activation
• When activation is greater than the threshold theta the perceptron outputs 1
• 
  

Data Science for Business by Provost Fawcett data science book review

• Interesting business approach to understand and apply data science
• Use verbose texts when explaining simple concepts
• Introduce advanced concepts like Information Gain and Entropy early in the book pg51. Less accessible for beginners. Good for people preparing for interviews and people who already got an introduction to machine learning
• Writes the full formula out, avoid using sigma notation to make it more accessible for business readers
• Useful graph of entropy pg 52
• Helpful graphical illustration pg54-pg55 two trees with different information gain
• The learning curve is not a slowly ascending one. The topics jump around in this book
• Variance measures impurity pg56

Subscribe to our mailing list

Email Address *

Udacity Machine Learning Nanodegree Linear Regression sample code

from sklearn import linear_model

reg = linear_model.LinearRegression()
reg.coef_
reg.intercept_

Codecademy SQL Table Transformation Subqueries Walkthrough

• 1. Table Transformation
• SELECT * FROM flights WHERE (SELECT code FROM airports WHERE elevation < 2000); 
• Nested subqueries 
• First get all the codes from airports table if its elevation is smaller than 2000
• Use this as a filter to query all columns of data from the flights table
• Don't forget the ; semicolon in the end