tensorflow beginner

http://blog.csdn.net/soulmeetliang/article/details/75666675


Knowing for beginners: https://www.zhihu.com/question/37180159
Author: Ni Feixiang
Link : https://www.zhihu.com /question/37180159/answer/96682815Source
: Zhihu The
copyright belongs to the author. For commercial reprints, please contact the author for authorization, and for non-commercial reprints, please indicate the source.

Summarize your own learning, come into contact with Numpy, Pandas, Matplotlib, Scipy, Scikit-learn, which can be regarded as an introduction, give your own trajectory (omitting installation), and summarize other people's answers, and finally there are Easter eggs. <img src="https://pic3.zhimg.com/50/8151389351191b7e1518ed30267e9e8f_hd.jpg" data-rawwidth="600" data-rawheight="85" class="origin_image zh-lightbox-thumb" width="600" data-original="https://pic3.zhimg.com/8151389351191b7e1518ed30267e9e8f_r.jpg">Numpy: To store and process large matrices, it is much more efficient than Python's own nested list structure, which is composed of C language development. This is a very basic extension, and the rest of the extensions are based on this. The data structure is ndarray, and there are generally three ways to create it. The conversion of Python objects is generated by numpy built-in functions like factory functions: np.arange, np.linspace.....read from disk, loadtxt Quickstart: Quickstart tutorialPandas:
A NumPy-based tool created to solve data analysis tasks. Pandas incorporates a large number of libraries and some standard data models, providing the tools needed to efficiently manipulate large datasets. Most statistical toolkit, better than R software in some ways. The data structure is a one-dimensional Series, a two-dimensional DataFrame (similar to a table in Excel or SQL, if you study deeply, you will find that Pandas and SQL have many similarities, such as the merge function), and a three-dimensional Panel (Pan(el) + da) (ta) + s, you know the origin of the name). What you need to master to learn Pandas is: summarizing and computing descriptive statistics, dealing with missing data, hierarchical indexing
Clean, Transform, Merge, Reshape, GroupBy Techniques
Date and Time Data Types and Tools (Date Handling Handy Flies) Quick Start: 10 Minutes to pandasMatplotlib: The most famous plotting system in Python, many others such as seaborn (for Pandas drawing) is also encapsulated by it. Creator John Hunter passed away in 2012. This drawing system is very complicated to operate, and it is prohibitive compared to R's ggplot and lattice drawing, which is why I personally don't abandon R, although
the graphics drawn by calling plt.style.use("ggplot") can roughly follow ggplot The color is displayed, but it still feels very tasteless. But the complexity of matplotlib gives it a lot of customization. It has an object-oriented approach and the classic high-level encapsulation of Pyplot. What you need to master are: the drawing of scatter charts, line charts, bar charts, histograms, pie charts, and box charts. The three major systems for drawing: pyplot, pylab (not recommended), object-oriented axis adjustment, adding text annotations, area filling, and special graphics patches. Students who use finance should pay attention to: You can directly call Yahoo financial data drawing (true ...) Pyplot Quick Start: Pyplot tutorialScipy: A
convenient, easy-to-use, Python toolkit designed for science and engineering. It includes statistics, optimization, integration, linear algebra modules, Fourier transforms, signal and image processing, often Differential equation solvers and more. It can basically replace Matlab, but it has little to do with data processing, and the mathematics department or engineering department uses it more. (slightly)
Recently, I found that there is a statsmodel that can complement scipy.stats, and time series supports perfect Scikit-learn: students who are concerned about machine learning can pay attention to it, a very popular open source machine learning tool, such as Google's open source TensorFlow at the end of last year, or Theano, Caffe (Jia Yangqing), Keras, etc., this is another problem. Homepage: An introduction to machine learning with scikit-learn<img src="https://pic3.zhimg.com/50/8151389351191b7e1518ed30267e9e8f_hd.jpg" data-rawwidth="600" data-rawheight="85" class="origin_image zh-lightbox-thumb" width="600" data-original="https://pic3.zhimg.com/8151389351191b7e1518ed30267e9e8f_r.jpg">Book: Founder of Pandas: Data Analysis with Python (Douban) (Recommended)
Textbook Collection of
: Scipy Lecture Notes