Hi, I seem to miss some basic ideas. I passed the test anyway but have a lack of understanding. To make it easier I reduced m from 30 to 5. (1.3 Dataset). These are the values:
Training dataset X:
[[-1.07296862 1.62434536 -0.52817175 0.86540763 -0.61175641]]
Training dataset Y
[[-62.17638819 22.06089266 -9.39343384 71.72956568 -45.984776 ]]
is X[0] and Y[0] the coordinate for a point in the plot? But the plot does not display this?
I have some more questions. But maybe they will be answered, when I get a better idea of what X and Y means. (Related to the scatter plot)
I know that I built the scalar product from every X with a changing W and b, which is different in every iteration and a cost function telling me, which values for W and b has the less costs. But at the end I am not quite sure what I want to predict (in this case)
Sorry. This feels all in all a bit embarrassing. Thank you for clarification.
Perhaps I am not understanding your question (and my apology if so) but your first question relates to X[0] and Y[0] if I am correct. This point is visible in the plot in the lower left corner, or were you hoping for the plot to display the values of X & Y on the chart? That’s a different challenge
What the algorithm is predicting is an approximate line that is as close as possible to every one of the points. It will never equal every point, but ideally the end result of your prediction will be a line that is closest. I hope I have understood the question.
We start the process by first training the model on many samples. A sample contains (x,y) pair of values → For a given value of x, it specifies the corresponding value of y. The model uses these samples to learn the relationship between X and Y, which is represented in the form of an equation y = wx+b. In the simplest case of a Linear Regression model, the relationship between X and Y will be in the form of a line. Once the final values of w and b are obtained, the training is complete. From there on, you can give any value of x as the input to the model and the model will output the value of y using the equation y = wx +b.
Thank your for your response. My question was indeed a bit vague. I mainly was confused about the scatterplot because I could not figure out how the X and Y values relate to each other. I also could not see the point in this prediction. But as @shanup says, it’s just to find out “W” and “b” to get from X to Y. And for whatever reason I lost this perspective. Have a nice weekend
