Machine Learning for the Non-Machine Learners: A Machine Learning Recipe with with Right Ingredients for New Chefs! (Part 3/6)

Welcome to another post where I discuss the third question (out of 6) to ask to qualify a business problem as a machine learning problem. Please refer to my earlier posts to review question 1 and question 2. Alternatively you can bookmark my blog and follow all the entries here.

Alors on danse! Believe it or not you can leverage ML to teach robots how to dance. The following video is made by a PhD researcher and lecturer of robotic and AI at University of Groningen. This researcher leverages the power of AI to teach a robot to dance in Persian style.

Want more? Watch this news piece in Dutch/English:

The good news for you as a leader is that you can find your way to dance between data science and machine learning and with that I introduce you to the third question:

Do you have data? I mean enough data?

That’s the question that you could be asking your team to start the conversations. But don’t forget that collecting data (good quality data!) is difficult. The key to the success of your algorithm is to feed it with good examples. The more the merrier! Without enough data and good examples the ML algorithm cannot produce reliable results. I won’t go into further details but to help you better lead your ML conversations I share two more pieces of information, after all we want you to sound as smart as possible to your team:)

The data or examples that I mentioned earlier should contain the following two items:

Attributes: characteristics of the examples/data

Label: the answer that you want to predict

For instance the CMO of Air Canada, could collect hundreds of thousands of Tweets on topics around vacation and summer planning and conduct a sentiment analysis to help Air Canada agents to develop better deals, reduce costs and improve forecasts all thanks to ML! If you are new to the concept of sentiment analysis you can visit Dr Yadav’s website for a quick practical review of what can be done with sentiment analysis. His research results on US Airline Sentiment Analysis is accessible here. If you are in the market for a sentiment analysis tool, drop me a line and I would be happy to help you with the assessment of the right tool for your enterprise needs.

Attributes and Labels are jargon that make sense to your data science team. They are the input/output that we briefly reviewed in the previous post. It is good for you to understand them so that you can connect the dots between a business problem and an ML problem. Like a well-orchestrated dance you will need to invest in a clear communication plan. You will need to lead your team to work in harmony. You should identify an ML worthy business problem and share it them with your data science team. The data science team builds and feeds the algorithm with quality examples and data. In the end, your organization manages to take a giant leap towards leveraging data as an strategic asset to increase the bottom line. The key to make this process successful is to ask the right questions at the right time.

If you think that your organization still qualifies to launch a machine learning project look out for my next post soon to learn what else you should be asking before you start your ML journey!

Like what you read? Give Zara Palevani a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.