Machine learning (ML) can provide unique analytical insights, as well as help automate some operational and decision-making processes more efficiently and effectively than non-ML alternatives. However, ML is also among the buzziest of buzzwords, and many are overselling and oversimplifying its usage.
Do not let anyone frame a data analysis, business problem, or process improvement as an ML use case. Instead, say: That is Not Machine Learning — that is a data analysis, business problem, or process improvement where ML might be able to help. But not before we evaluate other options. And with the understanding that ML is rarely going to be either the first or only aspect of the solution.
This episode is sponsored by: Vertica.com
Extended Show Notes: ocdqblog.com/dbp
Follow Jim Harris on Twitter: @ocdqblog
Email Jim Harris: ocdqblog.com/contact
Other ways to listen: bit.ly/listen-dbp
--------
26:12
Machine Learning is Label Making
Label Making. That is my simple two-word definition of Machine Learning. Machine Learning is Label Making. ML is LM.
Especially supervised machine learning, which creates either numerical labels (using regression algorithms) to make predictions about a continuous data value (such as sale or stock prices), or categorical labels (using classification algorithms) to assign data to pre-defined groups also called classes (such as Fraud or Not Fraud for financial transactions).
This episode is sponsored by: Vertica.com
Extended Show Notes: ocdqblog.com/dbp
Follow Jim Harris on Twitter: @ocdqblog
Email Jim Harris: ocdqblog.com/contact
Other ways to listen: bit.ly/listen-dbp
--------
15:00
Cloudy with a Chance of Data Analytics
Based on one of my presentations, this episode provides a five-part vendor-neutral framework for evaluating the critical capabilities of a cloud data analytics solution: Deploy, Store, Optimize, Analyze, Govern.
This episode is sponsored by: Vertica.com
Extended Show Notes: ocdqblog.com/dbp
Follow Jim Harris on Twitter: @ocdqblog
Email Jim Harris: ocdqblog.com/contact
Other ways to listen: bit.ly/listen-dbp
--------
27:36
Big Data Quality, Then and Now
A decade ago, just before the beginning of the data science hype cycle was the big data hype cycle. At that time I had the privilege of sitting down with Ph.D. Statistician Dr. Thomas C. Redman (aka the “Data Doc”).
We discussed whether data quality matters less in larger data sets, if statistical outliers represent business insights or data quality issues, statistical sampling errors versus measurement calibration errors, mistaking signal for noise (i.e., good data for bad data), and whether or not the principles and practices of true “data scientists” will truly be embraced by an organization’s business leaders.
This episode is an edited and slightly shortened version of that discussion, which even though it is from ten years ago, I think it still provides good insight into big data quality, then and now.
Extended Show Notes: ocdqblog.com/dbp
Follow Jim Harris on Twitter: @ocdqblog
Email Jim Harris: ocdqblog.com/contact
Other ways to listen: bit.ly/listen-dbp
--------
29:49
Three Questions for Data Analytics
Before you get started on any data analytics effort, you need to have at least preliminary answers to three questions: (1) What problem are we trying to solve?, (2) What data can we apply to that problem?, and (3) What analytical techniques can we apply to that data?
This episode is sponsored by: Vertica.com
Extended Show Notes: ocdqblog.com/dbp
Follow Jim Harris on Twitter: @ocdqblog
Email Jim Harris: ocdqblog.com/contact
Other ways to listen: bit.ly/listen-dbp
Data is often the basis for how we see the world, and how the world sees us. Understanding these data-based projections is the focus of this podcast, which discusses topics related to data analytics, machine learning, and data science. Produced and hosted by Jim Harris.
USA