Who

My name is Michael Lauren and I created this site to share my passion for machine learning, nlp, stocks & securities, cryptocurrencies, and data science in general. I am a senior software developer and data science enthusiast.

My appreciation for machine learning developed out of writing various types of neural networks in the domain of statistical arbitration to predict the prices of publicly traded securities. Taking this limited domain further, I became interested in the discipline of data science itself. I wanted to share some of this knowledge in a blog format, with code snippets and working repos. Occasionally, I will also add interactive examples of trained models with which users can interact.

What

DataDreams is, at the moment, a personal blog for my data dreams. One thing that I wanted to do from the get go, was offer a way for the average user to interact with trained (narrow) AIs. That's why I added a live sentiment analysis AI of the #Trump hashtag on twitter as well as a Markov Model Shakespearean sonnet generator that are both accessible in the more section at the top of the site. Whether you are for Trump or against, the conversation there is divisive, and it seemed a perfect testing ground sentiment analysis net.

I have been developing software and applications for over a decade and I have read my fair share of tutorials online. Online tutorials can be great when they are good, and abysmal when they are bad.

Sharing

Eventually, I would like to implement an API for various datasets I have collected while creating my blog posts. One idea I am interested in is facilitating users to help create datasets. For instance, the eventual goal for the #Trump hashtag Sentiment Analysis is to incorporate user feedback when the prediction is wrong to further train the AI model on the site, but to also offer the sentiment database collected on DataDreams for free through the API mentioned earlier. The Twitter sentiment corpus is quite large, but you will notice even on a decently trained model like the one you can interact with on the site, it fails quite often in emotionally charged political discourse.

That is just one example, but the idea stands. Data is hard to come by in this field, and helping other people get experience with real world datasets is one of the ideas I had in the back of my head while kicking the idea around of starting this site.