By understanding the cost function, to understand a few of the following points:
1. Univariate linear regression: h (x) = & + kx
2. Parameter: hypothesis
3. The cost function
4. modeling errors
This is h (x) univariate linear regression, as well as the cost function converts visual expression and modeling errors in the way we have seen is the minimum cost function J (x) is the lowest point of the function directly reflected in the two coordinate systems.
Of course, since these discrete sampling points, due to some circumstances, so we can not be artificial to sampling, not a way to solve the image, so we are adding more computer algorithms to guide the calculation and prediction.