Our server costs ~$56 per month to run. Please consider donating or becoming a Patron to help keep the site running. Help us gain new members by following us on Twitter and liking our page on Facebook!
Current time: 26th February 2017, 17:20

Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Machine Learning & Predictive Modelling
Machine Learning & Predictive Modelling
So over the past month or so, I have plunged into the field of Machine Learning and Predictive Modelling.  Data Science, and coding in general, to me is fascinating.  
I love that I can think up a crazy question today, scour the internet for relevant data sets, and pull that data together to see if I can't falsify my hypothesis.  Same thing with programming in general, I can think up a project that I want to do, and watch my project come to life without any real start up costs.  All the while I get to build on problem solving skills as I move forward.

Right now I just finished my first semi-serious predictive model(still have a bit more tweaking to do), and submitted it to kaggle.

I'm curious if there are any data scientists(professional or hobbyist) or kagglers out there that might want to provide me with some resources/tips/etc that helped them improve, or that might want to form a kaggle team(for fun, I'm not very good at this right now) and work on some projects together.

I little bit about my experience in CS:
I got started with CS in college, and took an internship as a DBA over at Verisign in the summer of 2012.  I liked coding a lot, but I hated CS in a corporate environment. 
While there I primarily worked with SQL and Java.
Afterwards I got into web development a little bit, and worked with Php(worst language ever), JS, and MySQL.  Around 2014 I essentially stopped coding and picked up some other hobbies.
At the beginning of 2016 I picked up python, and have been somewhat consistently(over the year probably an average of 6+ hours a week) coding in python since.
I'm familiar with the scipy package(Pandas, Numpy, sklearn, seaborn, etc)  
I have all the mathematics you would expect an engineer to have, and haven't really used any of it since college.  But I can grasp some of the Machine Learning fundamentals without needing to fill in many mathematical gaps.

Anyway, any Data Scientists(professional or hobbyist) out there?  Want to do some kaggle competitions with me?  Want to provide some helpful anecdotes, tips, papers, books.

I have one tip I think I can give that has helped me a bit, and I haven't seen many people using it to help them organize their data.
Most data sets will come with a .txt file that describes what each variable is.  Often times that documentation file will also in a fairly consistent manner explain what type of data you are looking at (ordinal, nominal, discrete, continuous), and you can use a RegEx to quickly automate splitting your data up into those types.  Spend a few minutes looking over the documentation, you should be doing this anyway, and try to see if you can't find a pattern that lets you use regular expressions to tease out important information like whether or not the data is nominal or discrete.

Kaggle Username: aristocatt
RE: Machine Learning & Predictive Modelling
Super interesting topic. I've used RegExes extensively for my last book, the LaTeX source for the entire 600 page thing was generated from input in different proprietary formats using RegExes in Perl. That's not machine learning, but a lot of fun nevertheless. I am a complete ML noob, I just started watching/reading "Learning from Data" by Abu-Mostafa last fall but haven't got the time to make progress quickly although he is a very good teacher.
The fool hath said in his heart, There is a God. They are corrupt, they have done abominable works, there is none that doeth good.
Psalm 14, KJV revised edition

RE: Machine Learning & Predictive Modelling
What's the book?

I am also definitely still a noob, but I am making a bit of progress.
You might be interested in Kaggle, it really gave me a nudge in the right direction.
I am the kind of person that enjoys seeing results as I build a working knowledge of a subject. And kaggle is a good source for that kind of stuff.
The forums there have lots of kernels that you can pull from and just build upon from other users. And they also have some quick tutorials that help get your feet wet quickly.

The ML packages are also great since you can apply them with a marginal understanding, and then build upon them as you research them in depth more.

Really it depends on how you feel rewarded when learning a subject. For me the application and results as I learn is a solid motivator.

I'll have to take a look at "Learning from Data".

For the sake of "this is not a kaggle advert", I am not at all affiliated with kaggle other than as a consumer of their platform.
RE: Machine Learning & Predictive Modelling
(9th January 2017, 14:43)Aristocatt Wrote: What's the book?
Quote:I'll have to take a look at "Learning from Data".


Very good, but relatively theory-oriented
The fool hath said in his heart, There is a God. They are corrupt, they have done abominable works, there is none that doeth good.
Psalm 14, KJV revised edition

RE: Machine Learning & Predictive Modelling
For anyone that is interested, this is a pretty fascinating paper/application of a neural network.


Original link found here:

The above link is a review of ML techniques for cyber security. And, not surprisingly, some of them also seem to be good hacking applications.

Possibly Related Threads...
Thread Author Replies Views Last Post
  telepathy machine reconstructs speech from brainwaves Jay1982 0 606 5th February 2012, 21:29
Last Post: Jay1982

Users browsing this thread: 1 Guest(s)