Introduction
This is my foray into understanding machine learning from a PM lens. As AI and ML become more mainstream, it’s essential as a PM to understand the nuances and more importantly, how to leverage it to make better products.
Organisations across the world are using machine learning and artificial intelligence to make their products do better- automate better, personalise better and predict better.
To get started, I will start from basics of ML and then deep dive into applications of ML in product management, machine learning algorithms, how to build models for different use-cases. This will be multi-part introductory series into these various aspects of machine learning.
Part 1: Basics of ML, examples, where do you use ML
Part 2: Types of machine learning, what is a machine learning model, example of a model, creating a model
What is machine learning?
Arthur Samuel, in 1959, first defined machine learning as the ability for a machine to learn without being explicitly programmed. What it means is that instead of giving computer set of inputs and set of rules to determine the output, we give the computer previous inputs and outputs, let it learn from those patterns and come up with an output for a new input on it’s own without being programmed.
For example, when you program a computer to say, identify whether the given image is an apple, you give the instructions or set of rules to the machine about the color, size, shape etc to say whether something is an apple, or an orange. However, in machine learning, what you do is you give the computer examples of what an apple looks like by feeding in hundreds or thousands of images. The machine then learns on its own, makes a model around it(set of rules and patterns) and when the new input comes, it can determine on its own whether the new image is an apple or not.
Difference between machine learning and artificial intelligence
A common question that is often asked is what is the difference between machine learning and Artificial Intelligence. Machine learning is a subset of Artificial Intelligence. Artificial intelligence as the name suggests is simulated human intelligence, which is much broader theme than just machine learning. Machine learning for example doesn’t understand context, or doesn’t provide solutions. It learns from previous inputs and outputs and predicts the output for a new input. So, it doesn’t understand the why or how of the relationship between the inputs and outputs whereas AI is becoming more and more capable of achieving that.
Where do you find machine learning in your day to day life
- Recommendations on online shopping sites, amazon.com based on your search, your previous buying history.
- Recommendations on spotify
- Automatic filtering of spam on Gmail
- Identifying fraud in financial systems
What ML can do, what ML can’t do
Famous Chinese philosopher Confucius said, “ to know what you know and what you don’t know is true knowledge “
We need to understand not only how ML works and how to apply it, but we also have to know when it works well(when we should apply it) and when it doesn’t work well(and when we shouldn’t apply it)
Let’s start with discussing what machine learning can do well. It works well if it’s given a sufficient quantity and quality of data. If we don’t have enough data or the data is not clean(we have a lot of outliers or missing data), we’re not going to be able to make machine learning work well
As stated before, machine learning can do following things particularly well
1. Making predictions by learning simple input output relationships. Examples of this might include predicting demand for a product based on things like time of day or season of year or temperature outside.
2. Personalisation for individual users. So if you’re a netflix subscriber, for example, you should be very familiar with this. Netflix provides personalized recommendations of movies to watch to users based on the machine learning model.
So now let’s talk about some of the things that machine learning cannot do very well.
- The first is understanding context. As humans, we can understand when a sentence is intended to be a joke or when someone says something that is intended to be sarcastic in nature versus taken literally. Machine learning models today are unable to understand things like jokes or sarcasm.
- Determining causation. Machine learning identifies patterns and correlations in data but does not determine cause or causation whether one thing caused another thing.
- Third thing, machine learning cannot do particularly well is explaining why things happen. So machine learning again identifies patterns but doesn’t attempt to explain why these patterns are occurring. It can explain outputs in terms of correlations with given input features. And again, it doesn’t explain why this combination of input features results in a certain output.
- And finally, finally, machine learning is not capable of determining the impact of what we call interventions or possible solutions on the problem. It’s also not capable of finding solutions to a given problem.
