Jack of All Data, Master of Some

The word sláinte in Scottish Gaelic means health. It is normally used in place of cheers. The word whiskey is actually a translation of a Gaelic phrase which means “water of life”. Another thing associated with Scottish life is bagpipes. I am appreciative of the time I have spent here as I listen to the not-so-distant bagpipist hike up the same hill as me. This last year has exposed me to more cultures and ideas than what I ever thought was possible. [Read More]

Clustering of similar product items can help decision makers strategically plan product mixes which can affect overall business strategy. In this post, I go over how to group similar products together using data from an online events retailer. The Technical Gist Don’t “blindly” standardize/normalize features without thinking about what you are actually trying to convey with each feature. Sometimes it makes more sense to normalize row-wise. If you value interpretability of clusters over all else, the features you engineer must make sense and come from hypotheses that you have formed from looking at the data. [Read More]

This post will deal with how to approach a problem using simulation. Furthermore, some theory on the distribution of order statistics shall be touched on as well as some very basic Queueing Theory concepts. Context Say you are trying to reach an international number. It can cost a fortune to reach an operator on the other side of the world. You have limited funds and you wish to simulate a call centre system to estimate how likely you are to reach an operator before you run out of money. [Read More]