The record about their own learning machine learning notes.
Today began studying "beginners learning machine learning Python Quick Start" This book is basically in front of more than 90 pages are talking about some of the basic grammar and some basic concepts of machine learning Python, so it is a quick browse over.
(The concept of tuples where there is a little bit deeper to understand is that tuples can not be modified list)
Machine Learning Highlights:
machine learning data generally have two groups, one group is a training data set is a data validation;
machine learning algorithms process:
1. Select the model function mx_fun, mx_fun machine learning function interface is our custom;
2, and the feature data set x_train training with corresponding features (results) the data set y_train, input model function mx_fun.
. 3, machine learning system built function, automatically analyzes the relationship between the characteristic data and the result data. Such a process is the machine learning process, but also algorithmic process modeling.
4, through machine learning and data analysis of training data, the system generates an AI machine learning models, we will save it to a variable mx.
5, the test data x_test input model variables mx, mx calls the built-in analysis functions predict, to produce the final results of the analysis y_pred.
6, if a firm offer, enter the most recent data, for example, today's stock market data or are selling Lottery game lost rate data, the system automatically generates a relevant forecast data, such as: day or the next few days the stock market trend data or the result of winning or losing the game.
Own understanding: this part a bit chaotic at first look at the time, and now his hand to knock again after this description find yourself with a new understanding of this process.
In their own words, you first need to have x_train and y_train training set, then the training set showed the model function (like on a linear regression function), after using the machine learning system function, according to the training data set and the specified model function to train learning outcomes with specific parameters can then be used to test data for this study good model for testing.
Here the use of a template class in C ++ again look at an analogy, is the equivalent of a model function template class, it is no entity can not create object. Our training goal is to train a template in accordance with the entity class out of the training is to achieve the basis for determining good training data.
Behind the experiment code yourself still do not understand, just knock shining again. But summed up, are all basically call the library, their main job is to train with the data handling and other direct use of library training tool can be achieved working model.
Here today to remember it!
