Pivoting content: arts and crafts, electronics, technology, and a little dose of everything else! Every Monday and Thursday (hopefully)!



Explore Data Visualization For Machine Learning

Star InactiveStar InactiveStar InactiveStar InactiveStar Inactive

Machine learning is a complicated subject and field to really get into and understand. However, we can find ways to understand and see how it works visually, which makes it fun and easy.

Disclaimer: I'm still a beginner in this field, therefore I may be incorrect.

What Data Should Be Visual:

Data is classified in a number of ways. We can record user interactions, save information from objects, or simply keep track of the number of books in a library (kind of a lame example). 

Here are some fine articles that discuss data in a more coherent manner.

Data Is What?

Beautiful Data

In those articles, which are pretty short, you will understand what exactly people mean by "those who have data are powerful" and etc. It's because the people with data can use it to draw conclusions about people and exploit them. However, that is unethical, and of course, I (or we) have good ethics and would not do that. But it is scary, isn't it? Literally, FaceBook, Google, or any other company essentially know your life, but this has been known for a decade now, so not really a surprise. If you still wonder which data is the most commonly used and collected, then it is actually personal data. Other types of data, like sensor and such, have value, but at this moment, personal data has the most value, especially when it comes to Unicorn companies, social media sites, etc.

My Idea On Visualizing Machine Learning Data:

First, machine learning is a subset of computer science, where you use algorithms to help the computer make better decisions, guesses, and prove solutions to the defined problem. If there is no problem to solve, then you do not need machine learning in your product or project. However, how would one want to better understand the steps and choices an algorithm would make? Even today, scientists wonder why does the machine think its answer is the best one. 

Machine Learning is alchemy

So really, it is a mystery. A computer is essentially more intelligent than us humans. In retrospect think of it like we are giving a computer a hammer and a manual, and then see what the computer comes up with. That's basically how we humans have come this far today. We learned from our mistakes (really big ones), and the computer does the same thing, but except 1000x faster.

Below is a video I made that is interesting and will go over most of my idea.



The picture that is very helpful for knowing most of the basic charts used for statistical analysis:


So this is what a spiral chart is explained as:

What is a spiral chart?

Here is a picture of what you can find from searching:


Of course, the "example text" and such would be replaced with actual classifications and be in 3D. Then you would be able to see all the outputs from the machine learning algorithm leading up to the best fit solution. So, the spiral chart above would also be inverted, going from bigger to smaller, from bottom to top.

Exploring Data Visualizations:

As I mentioned in the video, here are some links that we talked about:







The normal ways to visualize data is mostly by using languages like SQL, Python, Matlab, or R. However, you mostly used traditional charts for those. Although I'm not an expert in statistics and probability, I will learn about those and try my best to incorporate many techniques (algorithms) into my live projects.

There are many libraries for each programming language that have their own way of easily recording data and charting it, but if you want to start learning about it quickly, I highly suggest either Python or Java and Scala. Those languages have a huge community and support for machine learning, and you may learn fast!

Thank You:

I know there is more information that I left out, but I will add just a bit more after I come back. Thank you for reading and have a fantastic day!


Tags: Computer Science

Comments powered by CComment