Regression Explained: Predicting Continuous Values in Supervised Learning

Key Takeaways

Hey folks, in this video we'll discuss about regression. We know that regression is a type of supervised learning, right? Regression generally used to predict continuous values. Let me take a real life example to explain regression very clearly. A real estate company approached me and requested to create a model. This model's job is to predict site prices in Bangalore city. uh when a customer provides details like um area and square feet which they are looking for the model need to detect the prices and let the customers know. Now my job to create this model uh so first step I need to create a label data set. To create a label data set first step I need to collect the data. Uh what data I'll collect? I'll collect all the areas and locations where exactly sites are available. And then I'll get the square ft rate of that area because every area has different square ft rates. Right? Now I'll also get total area of the site like whether it is 30 40 or 60 depends on the site area and finally I'll get the price of the site. Now in this data set the price is our target variable because that is what we want the model to predict. You didn't get it right? Let's say now let's say uh a customer is asking to our model uh I need a site in JPEG 30 40 dimensions. Now our model need to predict the price of that site. So in our label data set the input features will be area square ft and total area of the site. Now our output will be the price. So output is also called target variable. Now if someone questioned you how exactly you can tell this model is supervised learning. Well we studied that supervised learning we give both input and correct output label to the model. Model learns from this label data. Then it predicts the output for new unseen inputs. Right? That's what we studied in our case. Price is label, price is a numerical value and price can increase or decrease. So then how can you tell this model is regression? Well, again uh the output is numerical here, right? The output is also continuous and the output is not a category. So this problem clearly falls under regression. So I hope you got about the regression in simple analogy. If you want to remember regression models, remember that u always your target or output value will be numerical. That's about regression. In coming classes, we'll discuss about classification and it types with real life examples. Hope you like this video. Uh see you back in coming videos. Thank you. Signing off mitten.