Lean principles in manufacturing are focused on reducing waste and lead times. Preventing unplanned equipment downtime to improve throughput is a key activity. There are two primary algorithms for analyzing equipment data - univariate and multivariate anomaly detection. Univariate anomaly detection focuses on analyzing the behavior of a single variable over time, for example, the temperature of a machine. It can be useful for detecting simple patterns of deviation from the normal behavior of a single variable and is relatively straightforward to implement and understand.

We expect our C-suite executives to make data driven decisions, but how much analysis should we expect them to do? Should a CEO or COO be using a dashboard or is that what others should use to answer their questions? I don't think many in the C-suite are "hands on" with data and I doubt that will change in near future. Some say It's better to to have them set and support a data-driven strategy. However, a data-savvy CEO would provide instant credibility to analytic initiatives.

Many companies are working on a digital transformation of their business. These expensive and time-consuming efforts are attempting to take advantage of new technologies like IIOT, 5G and artificial intelligence. My advice is to start with making all of the key business decisions using data-driven methods. The biggest challenges with accelerating data-driven decisions are prioritization and scalability. Set targets for the value of the automated decisions - not just a count.

How to understand the history of artificial intelligence in the popular press in five easy steps - 1. This technology is amazing! 2. We thought it was amazing, but it's actually terrible! We've moved on to something else. 5. Repeat. I've seen this for data mining, big data, machine learning and deep learning. What's the next AI technology that will be run through the cycle?

Over the last few months I've read more and more about ethical considerations in AI. A few big tech companies and consultants have created their own ethical frameworks, but have ethical considerations been implemented with real AI practitioners?

Why is there bias in AI? Bias in artificial intelligence reflects the environment where it was created. This environment includes the individual creating the algorithm, their experiences and background, the culture of the company/enterprise where they work, and ultimately the society where they live. One might also consider that, so far, artificial intelligence solutions are created by humans, which by our very nature have certain biases. So, given AI is created by humans and all humans have biases, should we give up hope for a bias-free AI? Probably. Should we give up hope to create an AI that makes better decisions than humans alone?

Full stack Data Scientists are a dying breed A decade ago all Data Scientists were full stack Data Scientists. The field was new and they needed to be able to find and clean data, develop analytical models and present their results without the assistance of a team. Since that time Data Science has grown significantly, both in terms of technical complexity (e.g., deep learning) as well as demand from industry, academia, and government. Similar to how physicians have become increasing specialized, Data Scientists are now part of a broader team that includes Data Engineers, Deployment Engineers and AI specialists. The days of a lone Data Scientist making significant contributions are done.

Look carefully at their business model., The primary limiting factor for AI algorithms is the amount and quality of the data. More and better data means better models. Thus, having substantial knowledge about the business domain reduces the time necessary to get good data. So if the startup is packed with AI specialists, but no domain-specific experts, be cautious.

Not many would have predicted the hype around these technologies in the last few years. But, given the time of year, I'm going to try and make some predictions for the 2018. My first prediction is that large companies will push for automated models to drive their critical business processes just like is currently being done in credit scoring and in direct marketing. This leads me to my second prediction. To enable this drive to automate, companies will need to scale their Data Science and Machine Learning efforts by developing specialized roles for data engineering, data science and model deployment.

I'm a fan of the a16z podcast and was intrigued with their recent discussion on the'end of theory.' The idea is that Artificial Intelligence tools, like unsupervised learning, don't require any hypotheses to generate insights from the data. From my perspective as an analytic executive in a large company, I'm going on the record that this is a terrible idea. While there are some startups that can embrace this'black-box' approach, the fast majority of large companies are run by people, not algorithms. These people are smart, and require evidence before they will turn over critical business functions to something they don't understand and can't explain.