While jumping on data science hype train few years ago I thought that it will be tremendous journey.

And it was, but in actual scheme of reality, things and feelings won’t last long.

I decided to list few things that makes me feel sick on my daily routine. And it really does not matter if it’s a spare time project or my actual job.

 

 

Extraordinary faith in data visualization and data analysis.

 

Data exist and is independent of who are looking at it. To not to be missunderstood, I really like to look at the data, perform tests, find new patterns. But data analysis did not stretch the data for your usability, it is meant to explore it, go deeper, find some connections. Data analysis is not intended to hurt itself and the origin.

Truth is not what you want it to be; it is what it is.

Lack of additive values which comes from deploying any dl/ml solutions.  

I think that it is not a secret that we are leaving in a bubble in which there is no real need for automatization at that scale.  

  • What has to be done, someone just has to calculate losses which are generated by pilot traps and not properly development projects.

  • I bet all of my saving that additive value is less significant that costs which were incurred.

 

Data Scientists become an API guys.  

Hype train pass by, now all you can do is just starting to retrain to software engineer. There are few signals which indicate that appetite for data science is not longer that big. At first, many companies started to realize that

  data scientist != data engineer*

and there is huge gap between creating a model and creating an effective solution. To clarify, I am not looking at it from FANG point of view.

Don’t get me wrong by calling Data scientist an API guys. What I mean is that

  • entry threshold for data science jobs become much smaller.
  • data science started to become glorified business intelligence.
  • automation and oss create an perspective in which knowledge of algorithms and measure is not needed with that much details.
  • I am grumpy guy in my 30s who just missed using AIC/BIC and statistical significance of each feature.

Awareness of Artificial Intelligence solution is still relatively low.  

I can list many situations in which expectations were much higher than what reality can create. But on the management and business stage there is more like ‘fake it till you make it’ point of view than ‘your arguments are right, we should stop doing this’.

 

I observed many situation in which misinformation and lack of proper communication appears. And it was usually determined by KPI indicators. If you ever plan to deploy AI/ML/Dl solutions whatever, and the only need to do that is to fill the gap in your team KPI, stop doing this right now. It is destructive not only for project but for people who develop that solution. Sooner or later people start to observe that additive value does not exist. They start to see that the only additive value from your hard work has an impact of how much of annual bonus you will receive.

It can just hurt people feelings and zeal for work, which can be an irreversible process.

I’ll expand that list, but I am now in brainfog state, so please be respectful.  

Sincerely, s3nh

All opinions are my own.