前面几章我们讨论了给定一些features，比如房屋面积，怎么通过TF的线性回归来预测结果，比如说房屋价格。下面会讨论logistic regression，它会通过输入的features来实现分类。比如说通过输入的图片，来进行0-9的分类。

经过前面介绍的单feature的线性模型，损失函数，梯度下降（part 1），epoch， learn-rate, 可变梯度下降(part 2)，我们可以继续介绍TF下面多特征(multi-feature)的线性回归。

上一章讨论了怎么在TF里面训练机器学习模型，以及基本的tensorflow代码。这为接下来讨论各种训练变化比如stochastic/mini-batch/batch, 以及adaptive learning rate gradient descent预铺了道路.

python的群里有人询问深度学习的基本概念，正好最近看到一篇非常简洁明了的介绍TF的基本概念的文章，尝试把它翻译成中文。以下为原文翻译：Tensorflow(TF)是google开源的深度学习框架。它有初级和高级两个教程实例。然后那两个例子把ML和TF放在一起解决一个多feature的问题--图像识别。本文尝试从最简单的线性模型来介绍一个单feature的问题，然后从这儿开始扩展下去。

整理了一下以前安装docker和tensorflow的东西

to avoid conflict, it is better to install tensorflow/theano in the virtuenv in python. Another way is to install them in the conda enviroment. This is the notes how to do that.

Two sigma provides the interesting data: y is a series of capped and floored time series which converged by time. The explainaroty variables have three types: fundmental, derived and techinical. This data also has a lot of missing values. All together makes the prediction interesting.

this is the exploratory analysis of the data in kaggle Two Sigma Connect: Rental Listing Inquiries. The data itself is very easy to understand. Here it focus on figureing out the relation between the explainatory variables and the dependend variable. Exploring the relation between x and y is very important in building a predictive and powerful model. This is the step one.

install docker, and dl-package including tensorflow, theano, keras and caffe on docker for dl class, set up port forwarding for remote ssh access on the new installed ubuntu 16.10

Ubuntu was install on vmware. More spaces are needed since lots of data download. This is a note how to increase the disk space: add new partition, set up file system, and mount to the dir.