Supervised Learning Supervised Learning
We algorithm to a data set, including the correct answer. Then the learning algorithm can give more correct answers.
This is also called regression. Return means that we try to predict the value of a continuous attribute.
Think of functional equations in mathematics.
Polynomial equations, how to get results from multiple dependent variables.
That can be found in the relationship between the data in some of the properties with the answers, we can launch a formula.
Classification means that we try to predict a discrete output value , there may be many types.
test
-
You have a lot of goods with a stock, assuming you have thousands of pieces to sell the same goods. You want to predict in the next three months, you can sell many pieces?
-
The second problem is that you have many users, you want to write a software to check each customer's account for each client's account to determine whether the account is compromised or destroyed.
Please judge regression or classification problems.
It is very simple. .
Unsupervised Learning Unsupervised Learning
To a data has not been informed of the results, I do not know what each data point, we just been told there is a set of data, you can find some kind of structure in which it?
For a given set of data, unsupervised learning may determine that the data sets contain different data clusters.
This is a clustering algorithm.
For example it is. News topics. How to put a lot of news to automatically classify and extract theme.
We did not know in advance the name of the theme of the news, the news is our theme for the data.
Analysis of social networks, how to analyze the same circle of friends.
Market segmentation and sales. Theory of galaxy formation.
Audio processing line of code:
[w,s,v] = svd((repmat(sum(x.*,1),size(x,1).*x)*x');
Andrew Ng describes a tool for language Octave.
to sum up
Some find the concept seems very simple look.
Found himself had intended to see, has been dragging did not see. .
