Regression is a statistical tool that helps us determine if an occurrence of one thing could be related to another.
Let’s say you run a furniture company. When you have no sales team, you have sales worth 1000$. When you hire 1 person for sales, you have sales worth 3500$. When you hire 2 people for sales, you have sales worth 6000$. How many sales people do you need to get sales worth 11,000$? Is there a relationship between the number of sales people and revenue?
This is a what is called a linear regression problem. As you can probably guess, the answer is 4. How? You need 2 things to figure out the answer.
1. How much sales you made without any sales guy. This is the amount you will make always. For our example, it is 1000$ & 2. How much sales does each sales guy get you additionally. For our example, as you can see it is (3500–1000) = 2500$
This means when you have 2 sales guys, you can get sales of 2500$ + 2500$ from them + 1000$ what you were getting without any sales guy. This gives us 6000$, as mentioned in the example. If we break down the above solution into an equation, we get:
Sales = 1000$ + no. of sales ppl x 2500$ OR
Y = a + b x X
Here Y and X are two variables, a is called the ‘intercept’ and b is the ‘slope’
This is linear regression! It represents a relationship between X and Y, which we can use to predict values of Y for a value of X we want.
But every correlation doesn’t imply causation! Among people who are interested in calligraphy, 71 percent said they would rather have loud farts than stinky farts [source]. But that doesn’t imply anything :D
— — — — — —
Get these daily blogs in your inbox. Subscribe now!
From the desk of Aditya Khanduri
About me (Why I started Polygyan)
