If you are in a leadership position the least that you need to know about machine learning is how to qualify your business problems as a machine learning problem. Therefore, you will need to know what machine learning could solve before you can decide whether it’s a worthwhile investment in your organization or not. You may say but Zara I am the HiPPO* I can hire smart people to run sophisticated algo for our company. In that case, I may ask whether you are doing that today.
If your answer is anything but yes please read on.
It is a challenge to keep up with the latest changes in Big Data, IoT, Blockchain, Artificial Intelligence, Machine Learning and who knows what the next big thing is. But the good news is that you can always ask the right question to find your way to the best solution. In this 6-part series, I will walk you through the machine learning questions to ask to ensure a clear line of sight to one thing that matters the most in business: profit!
Question 1: Do you really need machine learning?
Is automation an organizational challenge that keeps you up at night? Machine learning algorithms are supposed to transform a wide variety of industries. For instance, Google DeepMind saved Google 40% in data center cooling bills. You can read more about it here. Machine Learning can play a critical role in DevOps, Customer Service, Data Centers, Sentiment Analysis and more.
If you are dealing with a high volume of complex tasks then your company is a good candidate to use machine learning to save time and money.
In general, you have 4 categories of tasks:
On the small side
1. Small simple tasks (easy)
2. Small complex tasks:
Since you don't have to worry about the scale here, you can easily handle such tasks manually within a reasonable amount of time.
On the large side
3. Large simple tasks:
These tasks may look like good candidates but are not technically or financially justifiable to be handled by ML algos. After all, rule-based systems have proved to be operational for highly structured systems. Such tasks can be handled with rule-based algorithms as long as you know all the situations under which a decision should be made.
4. Large complex tasks:
Once your rule-based system hits a point where it needs to add handling various unstructured data sources such as JSON and/or XML to the current system, it's time to make a decision to introduce a machine learning practice to your organization. Such large complex tasks are excellent candidates for machine learning.
Please note that in your assessment, you should ensure to analyze the scale of the problem as well as the complexity of the rules.
So, ask yourself:
Do you need to automate large complex tasks at your organization?
If the answer is yes you may be off to an interesting journey. Please read my next post soon.
*HiPPO: Highest Paid Person in the Organization
#MachineLearning