In data science projects, different types of analyses are performed. These may include prediction, classification, pattern finding, forecasting, etc. You may also have to perform data clustering, sorting and trend analysis.
These analyses can be done in one of two ways. Either with traditional statistics or with machine learning techniques.
It’s important to realize that traditional statistics can move your analysis forward. You don’t always need to apply machine learning to every problem that you encounter.
Sometimes that is not obvious because everyone promotes AI and machine learning as the best data analysis option.
That’s often the case, but not always.
Remember, when you find an analysis that works, you also need to automate the process that goes with it.
When using traditional statistical techniques (like regression analysis or cluster analysis, for example), then automation can be done fast, it’s easier to explain and more transparent.
Of course, there could be excellent reasons to choose machine learning. But automation is also more complex and time-consuming, compared to statistical techniques.
So, a trade-off needs to be made at the beginning, when you are exploring types of analysis. You may ask, are traditional techniques enough for me to accomplish my project goals?
Because you need to make the right choices at the right time – and if you can avoid the cost associated with automating AI, why not?
Other video learnings:
Automating AI
AI is not plug and play
The process of decision making
Commonalities healthcare vs accountancy
