Module 1 - Machine Learning for Product Managers
We have started an online course to educate Product Managers/ Business Managers about Machine learning.
Subscribe to Youtube Channel — Technomanagers
What will you learn in this course?
In this course, we will tell you how to turn the business need into a machine learning problem, and then make a prototype out of it. If that performs well then how to deploy and productionise it.
This course is divided into 6 modules.
- In the first module, we will see, see the introduction to machine learning
- In the second module, we will see when to say yes or no to machine learning. Machine learning is very mainstream nowadays so it is very important for a manager to know when to say yes and when to say no.
- In the third module, we will see how to use machine learning weapons.
- In the fourth module, we will see how to prepare the training data for machine learning, that’s the first step in developing any machine learning model.
- From there we will move on to our fifth module where we will learn how to build the machine learning model
- Finally, in our last module, we will see how to deploy the model.
Module 1 — Machine Learning for Managers
What is Machine Learning?
I think every one of you has heard the term artificial intelligence and machine learning but how it is different from traditional programming.
In traditional programming, you have input and a set of instructions and with the help of it, you get the output. For example, if you have to do this calculation you can use traditional programming to do that.In machine learning, you have input and output but you need to come up with the set of instructions. Principally this is how machine learning works, this is not the exact way how it works but principally we can say that.
So what is machine learning, machine learning is basically Machine learning that works on input and output and comes up with patterns by the process of trial & error and learning.
6 Types of Machine Learning Problems
As a manager, you are trying to maximise the value you can deliver to the company, so before going into machine learning you need to understand the type of problem you are going to solve.
All in all your machine learning problem can be bucketed broadly into these 6 categories.
- Regression in which you are trying to predict numerical output value based on past historical data. For example, you have data on houses, their price and what attributes they have like locality, carpet area, amenities etc. Now if you want to know the price of a new house with different attributes you can use regression to predict that.
- The second type of problem is clustering in which you are putting similar things together. If you have shopped from Flipkart, their feature “Customer also bought X” is a classic example of clustering.
- The third type of problem is the recommendation problem which is simply recommending something based on your past behaviour. For example, Netflix recommends something based on your past watch history.
- The fourth type of problem is the Ranking problem. Google is using this to put the content on the top that you are looking for. This is a good machine learning problem because you can’t just use logic to rank this huge data.
- The fifth type of problem is the Classification problem, Gmail uses this to classify mail as ham or spam.
- The sixth type of problem is anomaly detection, for example, most trendy products on e-commerce or trendy tweets is an example of anomaly detection
3 Ways to solve Machine Learning Problems
Now that we have identified the broad problems, let’s see different ways to solve them. Broadly we can divide into 3 ways of solving any machine learning problems
- Supervised learning: In supervised learning, we have labelled data, labelled data mean we know what is input and corresponding correct output. We train our model and predict the output for a new input with the same underlying rules with which the model got trained with. Regression, classification, ranking and recommendation can be solved using supervised learning.
- Now comes unsupervised learning, In unsupervised learning, we simply feed the model with the data, not the labelled one. The model then discovers some patterns and labels. Clustering and anomaly detection can be solved by unsupervised learning.
- Reinforcement learning is an advanced form of machine learning. In reinforcement learning, the output is directly correlated with some rewards and punishment. For good output, the model is given some rewards and punishment is given for wrong or incorrect output. There are 3 major components states which are the current state of the model and actions what are the possible states the model can go in and then what outcomes it can receive going into those states.
Real-life examples of Machine learning
- Google Photos
- Facebook facial recognition
- Speech-to-text recognition
- Amazon “Recommended for you”
- Tesla’s autonomous cars
- Fraud detection in Financial Modeling
Comments
Post a Comment