A*******s
发帖数: 3942 | 1 最近一个面试肯定黄掉了，technical question回答得非常不好。有一类问题是，what
's the differences between SVM and Decision Tree? 觉得这类问题太广泛，不知道
该从哪里着手回答。亡羊补牢犹未为晚，我试着组织一下答案，大牛们看看应该从几方
面去比较两种predictive modeling：
1. scale sensitivity
2. problem of high-dimension data
3. computing intensity
4. incorporate nonlinearity
5. outliers, robustness
6. overfitting
还有么？ |
h***x
发帖数: 586 | 2 感觉你只要把SVM和Decision Tree怎么实现的用很简单的几句话说出来就可以了。
具体的细节他问什么答什么。你如果按你上面组织的答案去回答，反而显得你对这两种
方法都不是很精通。
Just my 2 cents,
what
【在 A*******s 的大作中提到】
: 最近一个面试肯定黄掉了，technical question回答得非常不好。有一类问题是，what
: 's the differences between SVM and Decision Tree? 觉得这类问题太广泛，不知道
: 该从哪里着手回答。亡羊补牢犹未为晚，我试着组织一下答案，大牛们看看应该从几方
: 面去比较两种predictive modeling：
: 1. scale sensitivity
: 2. problem of high-dimension data
: 3. computing intensity
: 4. incorporate nonlinearity
: 5. outliers, robustness
: 6. overfitting
|
A*******s
发帖数: 3942 | 3 有道理，不过三言两语能把一个方法说清楚也不是容易的事情，还得好好准备一下。
【在 h***x 的大作中提到】
: 感觉你只要把SVM和Decision Tree怎么实现的用很简单的几句话说出来就可以了。
: 具体的细节他问什么答什么。你如果按你上面组织的答案去回答，反而显得你对这两种
: 方法都不是很精通。
: Just my 2 cents,
:
: what
|
D******n
发帖数: 2836 | 4 u were answering as these two are same kind of methods. I dont know what the
interviewer wanted to know, but these two are totally different methods,
whats his point? as ls said, if he really asked this,then just answer how
they work and show they are totally different.
Unless, his questions is compare the pros and cons of SVM and dicision tree
or what's the difference in application and performance of SVM and DT.
what
【在 A*******s 的大作中提到】
: 最近一个面试肯定黄掉了，technical question回答得非常不好。有一类问题是，what
: 's the differences between SVM and Decision Tree? 觉得这类问题太广泛，不知道
: 该从哪里着手回答。亡羊补牢犹未为晚，我试着组织一下答案，大牛们看看应该从几方
: 面去比较两种predictive modeling：
: 1. scale sensitivity
: 2. problem of high-dimension data
: 3. computing intensity
: 4. incorporate nonlinearity
: 5. outliers, robustness
: 6. overfitting
|
A*******s
发帖数: 3942 | 5 i guess the interviews don't have much interviewing experience or they just
intend to test my communication. Some of their questions are unspecific like |
B****n
发帖数: 11290 | 6 at least you can compare these two methods on the classfication problems
the
tree
【在 D******n 的大作中提到】
: u were answering as these two are same kind of methods. I dont know what the
: interviewer wanted to know, but these two are totally different methods,
: whats his point? as ls said, if he really asked this,then just answer how
: they work and show they are totally different.
: Unless, his questions is compare the pros and cons of SVM and dicision tree
: or what's the difference in application and performance of SVM and DT.
:
: what
|
M*****t
发帖数: 85 | 7 对SVM几乎没什么印象。。。临时翻了一下才知道
lz这是面试的什么类型的职位，会被问predictive modeling里的SVM？是insurance
company的统计职位吗？
what
【在 A*******s 的大作中提到】
: 最近一个面试肯定黄掉了，technical question回答得非常不好。有一类问题是，what
: 's the differences between SVM and Decision Tree? 觉得这类问题太广泛，不知道
: 该从哪里着手回答。亡羊补牢犹未为晚，我试着组织一下答案，大牛们看看应该从几方
: 面去比较两种predictive modeling：
: 1. scale sensitivity
: 2. problem of high-dimension data
: 3. computing intensity
: 4. incorporate nonlinearity
: 5. outliers, robustness
: 6. overfitting
|
A*******s
发帖数: 3942 | 8 这就是我最郁闷的地方了。这个职位是关于BI软件的测试，我还以为看重的是
programing。结果俩面试官一上来就说，since you have adv sas certification, we
have no doubt about your programming skills. Let's talk about you data
mining experience. 晕死，我就完全没准备. 结果ANN，DT和SVM问了个遍。还问我能
不能在data step里实现SVM，我只好很老实地说我很多年没碰过quadratic
optimization了，完全想不起来细节了。
这个是有史以来打击最大的面试，让我觉得自己啥都不懂，唉...
【在 M*****t 的大作中提到】
: 对SVM几乎没什么印象。。。临时翻了一下才知道
: lz这是面试的什么类型的职位，会被问predictive modeling里的SVM？是insurance
: company的统计职位吗？
:
: what
|
D******n
发帖数: 2836 | 9 我觉得是他们公司想搞这个所以随便问问，他们自己也没有答案的。。
we
【在 A*******s 的大作中提到】
: 这就是我最郁闷的地方了。这个职位是关于BI软件的测试，我还以为看重的是
: programing。结果俩面试官一上来就说，since you have adv sas certification, we
: have no doubt about your programming skills. Let's talk about you data
: mining experience. 晕死，我就完全没准备. 结果ANN，DT和SVM问了个遍。还问我能
: 不能在data step里实现SVM，我只好很老实地说我很多年没碰过quadratic
: optimization了，完全想不起来细节了。
: 这个是有史以来打击最大的面试，让我觉得自己啥都不懂，唉...
|
