Relationship between Machine learning, Statistics, and Data Science

Question

Proposal: Data Science

I would like to know the how is machine learning used in data science. Statistics models are mostly used for smaller data set, e.g., hypothesis testing, confidence interval, etc. These models are broken in the so called big data world. Is statistics still useful? If so, in what way? Also Machine Learning, although being built upon statistical models, can offer predictions or results without going through the process of "human understanding". Does this indicate a bad future?

Anyway, I use Stack Overflow extensively. However, the programming or the computer science forums are very rigorously about the questions. In programming forum, you must ask a technical issue, and in academia forum, you can never ask an implementation problem.

It would be cool if this Data Science forum can provide a small niche space for people to discuss some philosophical problems, instead of focusing rigoriously on "problem solving". Any thoughts?

Answer 1

We have been experimenting and exploring a bit what what types content Stack Exchange can handle beyond the traditional Q&A. But there comes a point where the engine itself gets in the way of promoting useful discussion and can actually hinder it (this is, in part, by design and why we create this "knowledge reference system" in the first place).

The core engine of Stack Exchange is designed to host a type of technical writing that is more like an article in Wikipedia than something you might see in a traditional discussion forum. The questions and answer you provide here are supposed to be a work of reference, not a collection of conversations and debates. As such, Stack Exchange simply may not be well suited to "discuss some philosophical problems."

There's nothing inherently wrong with these questions, it's just that the Stack Exchange engine was not design for the type of back-and-forth conversation that these subject deserve. We never intended a Stack Exchange site to be the be-all and end-all destination for all conversations in that subject space. I can't really answer your question this early in the site development, but if these "conversation-starter debates" create a frustrating experience for the participants, sometimes we simply have to forgo them entirely.