在Coursera学习机器学习课程时，自己用python从原理上实现的各种机器学习基础内容.zip

# Machine-Learning-Codes-during-Learning I'm learning ML and here are some codes written during this time. (Coursera's courses are just AWESOME!!!!) 1 Linear Regression One can use this code to see how a basic Linear Regression method simulate the math function you gave.(Such as ordinary ax+b,ax^2+bx+c,sinx/cosx and so on).I also visualized the variation process of machine's parameters during learning to make it clearer.(Surely this makes the traning time much much longer than when there're no visualization) 2 Logistic Regression In fact I finished this a few days ago and have few to say, but logistic regression seems to be an important prework for Neural Network. They share the same cost function. Logistic regression also introduced the idea of regularization. For the code here, the Logstic_regression.py is a round regression of two-type and the second one's job is to seperate two type of data with a straight line. Visualized like below: (Ok, I didn't know that ReadMe can't contain Pictures) Well, just double click the file and you'll see what the codes do. 3 Neural Network Hello World It's a Three layer(One input, one hidden and one output) full-connected network that learns the handwriting numbers with MNIST's training datas. And the NeuralNetwork_predict.py is using MNIST's test datas to show the accuracy of the traning work. LearnedThetas contains nearly initial trained parameters that means it works with the network but its accuracy is not high. Also I wrote the LoadData.py which can load data from original MNIST files. I do think it's of much importance for this part.