Modeling a World of Probabilities
<p>In my last article, I explained regression where we create a hypothesis function g and tune the weights using SGD to get g(x, w*) which can output predictions for novel x values. However, we live in a probabilistic world.</p>
<p>If we examine the square footage vs house price example from the last article, we will realize that given the same square footage, the price of the property can vary, even if all other features are similar. This is because the price of the property lives along a distribution as opposed to being a set value. This means to make our models more accurate we should return a probabilistic distribution as opposed to a singular value. This would allow us to return that the price of the house should be y ± some range.</p>
<p><a href="https://medium.com/@mail.fyiu/modeling-a-world-of-probabilities-52fe2b9363fe"><strong>Click Here</strong></a></p>