zotroneneis / machine_learning_basics Goto Github PK

]

In the numerator, I think it should be $p(y|x, \theta)p(x|\theta)p(\theta)$, so why $p(x|\theta)$ disappear ?

I tested the exact code on collaboratory and found the following:
Test accuracy with k = 1: 97.77777777777777
Test accuracy with k = 5: 97.55555555555556

very good.

in chapter "linear-regression":

"n_iters" is omitted in section "training with gradient descent";

"c='orange'" is omitted in section "visualize test predictions";

in chapter "logistic-regression":

"* 100" is omitted in section "testing the model";

in chapter "k-nearest-neighbor":

"regregression" may be typo at the start;

"plt.show()" is omitted in section "dataset";

"test accuracy with k = 8" and "*8-like" characters may be typos, must be 4;

in chapter "simple neural net":

"\textit{sigmoid}" may be typo in section "forward pass";

in chapter "softmax regression":

"one-hot encoded" in section "step iii", confused by "So $y_k^{(i)}$ is $1$ is the target class for $\boldsymbol{x}^{(i)}$ is k, otherwise $y_k^{(i)}$ is $0$.", what is the meaning of "is 1 is the target...";

continuous updating...

Hi, what is the license of this work? Thanks a lot for sharing :)

(if you're not familiar with choosing licenses, Github has put up a website to pick one in addition to their page about licensing a repo.)

Thanks for these clear tutorials for basics of machine learning.

I have read through with great honor and translated into Chinese, and errata #7 as well.

Now i wish to contribute them as a Chinese version, just put this link into README.md can come true.

I think you should also mention about data preprocessing. It is one of the important and basic step of machine learning.

I had a tough time understanding LSTM completely and I believe there are many out there who may have doubts on it. It would be awesome to study LSTM (or any recurrent NN) with plain Python.

Can I work on Random Forest Classifier issue?

Hi,
I am new to machine learning. I already know advanced python. Currently, I am using anaconda spyder for my codes, but I was hoping somebody could recommend an editor solely for machine learning and focused on "pandas" as well as "NumPy".

Would be nice to see a naive bayes implementation.