May 8th 2019,
So, I have been off the interwebs for a little bit. I had been working with some really amazing people on some interesting project which I hope to share with everyone soon. In the meantime, I had been struggling with how to move forward. So far, this year I have focused on strengthening my theoretical background in deep learning. I have done this by taking the Coursera Mathematics for Machine Learning Course (Linear Algebra). It hasn’t been the most approachable course though. The course is not very well structured so I have been trying to use Khan Academy and Gilbert Strang’s Linear Algebra textbook to buttress my understanding of the concepts being taught. I hope to finish this course in a week or two and then move onto their multivariate calculus course as a move ahead.
I have also worked on my SQL skills quite a bit. I used MySQL to build a personal project of sorts for myself. Because I read quite a few books, I wanted to keep a record of them and my feelings about them. I wanted to formalize the list of books which I have read. I used Python and MySQL to build the database. It is a simple database, but one that can be scaled up. I have some ideas that I want to work on but I haven’t fully formulated them yet. However, I am in no rush to finish them.
To build this database, I also had to work with Google Sheet’s API which was quite a pain. I had records of some of the old books I had read as well as my thoughts and such in a sheet and I wanted to automate the import of all that data into my database. After some tinkering, I was able to get it to work. I have the code for it in this GitHub repository. This repository also contains a Notebook detailing my work. The code for my SQL database is here.
Also, I am going to be taking an active part in Kaggle competitions and building end to end pipelines for them. In keeping with this idea, I did my first competition in a while, the advanced regression for predicting house prices.
I know, I know. Don’t roll your eyes!! I did it mostly to warm up my feature engineering muscles! I was actually quite happy with the feature engineering I did here. Because of the feature engineering, my very bootstrapped simple regression model jumped to the top 66%. This was better than my other attempts at other competitions. I will tinker with it as I go and learn better things. Here is the link to my Kaggle Notebook.
Now, a little bit of housekeeping!!
I had been away from writing on a regular basis and I missed it. I will be writing on a more regular basis that I have been in the last few months. It won’t be every day as was the case with my 100 Days of Machine Learning blog. However, I think that once a week blog would be more realistic detailing the things I had done that week. So here is to week 1 and many more!!!