Logistic Regression Cost Function

[jjstuart79]

Hi,
I am studying logistic regression and gradient ascent and have seen it used with a cost function and without one. Could anyone tell me why you would use a cost function? It seems just as effective without one.

alpha = .05

h = data * weights
error = labels - sigmoid(h)
weights = weights + alpha * data * error

At this point I would think I can just loop through the above code and break when "error" converges. However, I've also seen code where you would pass "weights" to a cost function and break when the cost function converges. Please let me know if you need more information from me. Thanks

 

[Valkarie]

.A cost function is used to measure the performance of the model. The cost function is typically a measure of the difference between the predicted output and the actual output. The cost function is minimized in order to optimize the model's performance. In the case of logistic regression, the cost function would be the log-likelihood of the data given the weights. By minimizing this cost function, the weights are being adjusted so that the output of the model is as close as possible to the actual output. So, using a cost function allows you to optimize the model's performance and better achieve the desired result.