kaggle.com上的数据集有时候会比较大 ，而且没有提供网盘下载机制，国内下载速度非常慢，同时下载需要验证，也无法使用迅雷工具下载。
kaggle论坛上看到有wget的下载方式介绍[1]：
做法是先登录kaggle.com，记下浏览器中的cookie，将cookie保存到cookies.txt中，执行如下命令：
wget -x --load-cookies cookies.txt -P data -nH --cut-dirs=5 http://www.kaggle.com/c/avazu-ctr-prediction/download/test.gz
但是很快就执行完毕，只下载了14kb，肯定有问题：
[zhf@localhost ~]$ wget -x --load-cookies cookies.txt https://www.kaggle.com/c/avazu-ctr-prediction/download/test.gz
--2015-11-02 23:35:29-- https://www.kaggle.com/c/avazu-ctr-prediction/download/test.gz
Resolving www.kaggle.com (www.kaggle.com)... 168.62.224.124
Connecting to www.kaggle.com (www.kaggle.com)|168.62.224.124|:443... connected.
HTTP request sent, awaiting response... 302 Found
Location: /account/login?ReturnUrl=%2fc%2favazu-ctr-prediction%2fdownload%2ftest.gz [following]
--2015-11-02 23:35:32-- https://www.kaggle.com/account/login?ReturnUrl=%2fc%2favazu-ctr-prediction%2fdownload%2ftest.gz
Reusing existing connection to www.kaggle.com:443.
HTTP request sent, awaiting response... 200 OK
Length: 14687 (14K) [text/html]
Saving to: ‘www.kaggle.com/c/avazu-ctr-prediction/download/test.gz’
100%[===========================================================================================>] 14,687 --.-K/s in 0.03s
2015-11-02 23:35:33 (450 KB/s) - ‘www.kaggle.com/c/avazu-ctr-prediction/download/test.gz’ saved [14687/14687]
由上面的日志，可见，被重定向了到“ https://www.kaggle.com/account/login?ReturnUrl=%2fc%2favazu-ctr-prediction%2fdownload%2ftest.gz”去了。
于是，我们用wget的post数据参数提交用户名、密码。
[zhf@localhost ~]$ wget https://www.kaggle.com/account/login?ReturnUrl=%2fc%2favazu-ctr-prediction%2fdownload%2ftest.gz --post-data 'username=login_name&password=login_password'
--2015-11-02 23:37:18-- https://www.kaggle.com/account/login?ReturnUrl=%2fc%2favazu-ctr-prediction%2fdownload%2ftest.gz
Resolving www.kaggle.com (www.kaggle.com)... 168.62.224.124
Connecting to www.kaggle.com (www.kaggle.com)|168.62.224.124|:443... connected.
HTTP request sent, awaiting response... 302 Found
Location: /c/avazu-ctr-prediction/download/test.gz [following]
--2015-11-02 23:37:19-- https://www.kaggle.com/c/avazu-ctr-prediction/download/test.gz
Reusing existing connection to www.kaggle.com:443.
HTTP request sent, awaiting response... 302 Found
Location: https://kaggle2.blob.core.windows.net/competitions-data/kaggle/4120/test.gz?sv=2012-02-12&se=2015-11-05T07%3A39%3A03Z&sr=b&sp=r&sig=rKgKT2uZE6B4sLTirB1qdR8o262a9BgQPh233olSedg%3D [following]
--2015-11-02 23:37:20-- https://kaggle2.blob.core.windows.net/competitions-data/kaggle/4120/test.gz?sv=2012-02-12&se=2015-11-05T07%3A39%3A03Z&sr=b&sp=r&sig=rKgKT2uZE6B4sLTirB1qdR8o262a9BgQPh233olSedg%3D
Resolving kaggle2.blob.core.windows.net (kaggle2.blob.core.windows.net)... 23.98.55.152
Connecting to kaggle2.blob.core.windows.net (kaggle2.blob.core.windows.net)|23.98.55.152|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 123803952 (118M) [application/x-gzip]
Saving to: ‘login?ReturnUrl=%2Fc%2Favazu-ctr-prediction%2Fdownload%2Ftest.gz’
7% [======> ] 9,773,056 28.2KB/s eta 36m 24s^C
这样，虽然下载速度慢，但是可以放到后台去执行。
参考：
[1] https://www.kaggle.com/forums/f/15/kaggle-forum/t/6604/downloading-data-via-command-line