posts

Having spent the last year trying to get my head around machine learning I starting having a look at Blockchain & Bitcoin. Primarily I want to understand the technology enough to know which companies/use cases are fluff and which may have legs.

Dad works in the hospital and approached me with a headache concerning how to calculate the capacity of a room given its dimensions and a requirement of 2 meter socially distancing.

Like everyone else I have had six months of Covid imposed solitude but unlike others I’ve been extremely unemployed.

BJJ is my drug of choice and while I can’t train in lockdown I’m watching alot of old match videos. I found myself looking for results data from the most recent European Championships in the brown belt division to no avail. The IBJJF, the federation that run the major tournaments, release details of who medaled by belt & devision but provide no match level results. I decided to have a go at trying to programmatically piece results together.

After my original post on logistic regression I thought I could quickly cover model optimisation. This was very short sighted as there is too much to cover so I’m pivoting away from that idea. Instead I will post a series of notes covering concepts along the way to hopefully improving model performance on the Kaggle Titanic dataset. This post is going to cover accessing performance with a binary classifier.

This post is a learning resource to compliment Coursera’s Machine Learning course which I recently completed. I’m going to manually fit a logistic regression to get my head around the inner workings. I’m going to use the Titanic dataset from Kaggle to make predictions about which passengers survived the shipwreck.

As part of their transparency agenda Facebook launched their Ad Library reporting tool in March 2019 which gives users a view on ads relating to political and social issues in their country. Ads data in the repository starts from May 2018 and since that date you can query ads currently being run as well as those no longer active. In addition to the reporting tool they have an API for the Ad Library . You’ll need register as a developer and they send you out a code in the post prior to granting an access token.