A very common problem I find in the industry is senior managers not fully understanding how best to utilize Machine Learning technologies to help their business. A common misconception is that these projects should be treated the same as traditional analytics and whilst they can serve that purpose I think that is leaving most of the potential value on the table.
I have found that the best way to work with ML is to simply ask strategy questions and totally ignore technology and infrastructure. Questions such as:
- How can I reduce manufacturing costs?
- How can I increase productivity of sales reps?
- How can I make my call centre more efficient?
Once you have these “big ticket” ideas/questions then you start breaking this down into technical actions.
For instance, if we are looking at making a call centre more efficient we may start by breaking that down into multiple steps. Such as:
- Can I predict the number of calls I will receive on a specific day one week from now
- Can I predict which rep should take which calls
- Can I see what makes one rep more efficient than another
Any of these are examples of a perfect machine learning proof of concept. In my experience these proof of concept projects take a matter of days (at most a few weeks) and after that time you know whether you can answer that question using Machine Learning, with what kind of accuracy you can answer that question or at worst you are left with the knowledge of the information you will need to collect to revisit this question in the future.
So again, never start thinking about the data, about data warehouses, infrastructure, etc. Start with high level strategy, break this down into manageable questions and tackle them in a quick proof of concept.
How do you go about tackling one of these questions you ask? Well simple:
- Collect relevant data
- This could be an Excel export from a rostering system for instance
- This could be a database connection to the financial system
- This could even be free / purchased information from a third party
- Merge relevant information from this data into an experimentation dataset
- Rinse and Repeat
- Try new data, try new algorithms, try new objectives, etc, etc.
This process is quick, efficient and very effective way of using this amazing technology to quickly realize value for your organisation. The examples above are over simplified but this approach should be the starting point to any discussion into bringing machine learning into your organisation.