HN user
Sort of related, but my epileptic dog has most of his seizures right as he's waking up. Perhaps he's suddenly losing his "protection" as his short waves are disabling.
Hi - I submitted my application to the DS role, but am interested in the applied scientist role as well!
I love Pillars of Earth too, and he has turned it into a series (the Kingbridge series) with the fifth book expected this year. I've read each book and overall enjoyed each, particular his detailed research attempting to accurately depict everyday life for each historical era, however it does feel like the same plot with new characters 200 years later with each release.
I have used airflow with two different organizations over the past couple years. When we had a complex orchestration with critical pipelines and enough human-power to manage the system, it was great. Trying to deploy it for a small team with no critical pipelines has been overkill and we recently migrated to dagster which is still in beta but accomplishes 90% of what airflow does with a much smaller footprint.
And the esoteric notation!
The fundamentals of actuarial science: probability, time-value of money, and tail-risk have absolutely benefited me. I do gripe that much of the material is out-dated. I sat for all my exams within the past 15 years and (for example) we had a lot of option pricing models that have been archaic since the 90's and no machine learning (material has been updated with ML, but it constantly feels a couple decades behind).
One of the most frustrating aspects of the process is that once outside of insurance/insurance-adjacent industries, the credentials are useless. I have my FSA (fellowship of the SOA) and CERA (chartered enterprise risk analyst) from the SOA. I no longer wanted to work in insurance and left for data science. I felt the need to obtain a graduate degree to feel qualified enough for the data science space. My Master's degree was a cake-walk compared to the actuarial process but that is what my peers care about, and I always feel not quite at the same level as my peers with PhD's.
It's even more constrained than my experience. The U.S has two actuarial governing bodies: the CAS and SOA - the former handles P&C, the latter just about everything else (life, health, retirement). The two organizations hate each other and it's almost impossible to get a respectable job in one industry if you have the wrong credentials. What makes it even more crazy is that the two orgs have the same exact first couple of qualifying exams.
My advice has always been: only consider dedicating your time to these exams if you're highly certain this is the career you want. The pay isn't as great as it once was, you're constrained to legacy industries, the process is as time-consuming as a PhD without getting the respect that comes along with it, and you need to decide early on which area of insurance you'll want to practice.
I just checked out their website, it's really annoying. Zero information on pricing, you need to create an account to "try for free" to access that information.
I had experience developing automated underwriting models in the insurance industry. As the models are becoming more sophisticated and adapting machine learning, heavier scrutiny is coming from regulators.
For good reasons, guardrails are required to not only protect against discrimination, but also so proxies can't be used either. For example, we can't include race in health and life underwriting decisions. But since zip code is highly predictive of race, that attribute must also be excluded.
I'm not familiar with banking regulations, but imagine similar policies are applied. In these cases, being able to demonstrate that a model isn't discriminating is not only ethically important, but in many cases is legally required.
Auto-encoders have been more successful in fraud and anomaly detection then supervised methods. For the uninitiated: the basic concept is to reduce the feature space (i.e. the things you know) to a lower dimensional space, then decode back into the original space. When enough differences arise between the original and reconstructed variables, the event may be flagged for a human to review (or some triage process).
And the history of every life insurance application an individual filled is collected and shared through a company called MiB[0] to help mitigate fraud.
[0]. mib.com
Speculators have shorter windows than retail investors looking to store value for long-terms gains. Owning a non-dividend stock in my retirement fund is not equivalent to speculation.
I have been using ProtonMail for a year now and am very happy with the service.
Thanks for the suggestion. PopOS was not yet created when I purchased mine, it came with Ubuntu 16.04LTS. I used Ubuntu up until a couple months ago and decided to make the switch instead of upgrading to 18.04. Honestly I think my battery life has improved slightly but I haven't performed any benchmarking and my evaluation may be biased.
I purchased one about 2 years ago. I am generally happy with purchase, but certainly don't feel like I saved any money. I did need to reinstall the operating system (and have since switched to Manjaro). The fan is loud and battery is lousy, but the hardware has not deteriorated other than the connection to the monitor needed tightening. They also don't void the warranty if you open up the machine and tinker, which was a bonus because I wanted to install my own second SSD. I also really like the look and the keyboard is excellent. Plus a good number of ports including usb-c and Ethernet which wasn't as common two years ago.
Agreed, it is a little confusing. What makes it even more confusing is bias can be introduced as a regularization technique (to address overfitting) and these are not the same thing.
Bias in this context can be simply interpreted as analogous to the intercept in y = mx + b. Kind of like the baseline when no additional information is given.
I haven't read the article thoroughly but I suspect part of the data preprocessing step included centering and standardizing. Those must be the mean weight and height values.