发信人: sunfaquir (非礼勿言), 信区: JobHunting
标 题: 发几个面经(5) Groupon 电面+onsite
发信站: BBS 未名空间站 (Tue May 28 18:57:07 2013, 美东)
尽管第一个onsite是twitter给的
groupon却是我第一个去onsite的公司 过程也是所有面试的公司中最漫长的
面完groupon心态彻底变平和了 再面任何其他公司也都不会再觉得折腾了
总共面了 10个人 onsite前两轮电话+ onsite 5个人 + onsite后3轮电话
面试的过程中要去的组因为内部人re-org被强塞了几个人 offer自然也没了
电面
p1 主要面data mining,毕竟宽泛,考察到了
1) measures of classification
2) boundary decision for classification
3) Feature selection
4)Entrophy,TF,IDF
5) coding 给定query 打印出所有match的combination
// Query = dress for less
// Expansion: "dress:[es, ed, ing] for less:(cheap, deal)"
/*
dress for less
dress for cheap
dress for deal
dresses for less
*/
p2
coding题目 Print Binary Tree in Zigzag order
onsite
p1:
1) team introduction/self introducaton/project introduction
2) Coding: Given a user list and a deal, also a api (float relevance(User u,
deal d)), return the k top users that are most related to the deal
p2:
1) indroduction
2) 扔硬币:两个不知道的Head/Tail概率的硬币,扔1000次得500次head VS 扔100次得50
次head,算confidence, p value
3) Given a friend network:
Lan---> micheal
Michael--> Kathy
Lan--->alex
....
Found the following result:
number of person number of friends
3 10
2 4
...
a)how to do this? no need to code.
b)how to do this with Map Reduce
c)how to do this in SQL(count,group by, union)
4) Kmeans的 cost function
p3 lunch guy 算法 Model 设计题
Given a user query(String), how to build a classified model in which it
takes the query and return the most related deal categories?
Training data availabe(Deals info--->categorical info)
p4
1)聊天 project/team
2)coding 题目
Given two set of weather Data
April
City Tempratature huminity
San Jose 50 50
San Fran 40 30
May
San Jose 70 30
Chicago 30 20
Ouput the variation in the following format
San Jose 40% -40%
San Fran infiniti infiniti
Chicago -infiniti -infiniti
p5
1)问project 穿插考察概念
2) unfair 硬币问题
throw 1000 times, 550 times head, what's the probability of p=0.5?
3)算法
give two Arrays A and B (size large)
output T if all element in A are also present in B, otherwise F.
有follow up
4)算法设计题 一堆groupon向google买的广告的点击和收费记录
如何设计算法获取单个词条的收费rate
Problem
words # of clicks TotalPaid Amt to google each day ,
w1, 320
w2, 250
w3, 5
w4, 230
......
how to get the rate for each word?
提出解法后有follow up questions.
onstie后recruiter反应feedback很好 要时间说Hiring manager要电话讨论role
availability,放松警惕了随便给了个时间也没准备电脑 网络 耳机什么的 结果HM打来
电话是技术面,让coding的时候一下慌了 没面好,于是给recruiter发信解释,又加了
两轮面试(recruiter说是一轮技术面,一轮介绍产品的,后来证明又是两轮技术面,
groupon的内部沟通确实有些问题) 虽然后面coding写的都还可以 但这个过程中其实
position已经没了
onsite后 电面
p1
1) design question, big data/hash table related
2) coding
Given
"ABBEEFG" and pattern "BE"
a) remove pattern from string
return "ABEFG"
b) recursive remove pattern
return "AFG"
p2
dp题,给定钱数和所有的可能硬币面值,求最少的硬币数目的组合
p3
问了很多personal questions 略过不表
技术有关题目
1)How to improve an existing algorithm?
follow up给出了目前groupon 推荐deal的算法,如何优化
2)coding题目:给出二叉树,打印从 root到leaf的所有可能路径
--
※ 修改:·sunfaquir 於 May 28 19:35:00 2013 修改本文·[FROM: 98.]
※ 来源:·WWW 未名空间站 海外: mitbbs.com 中国: mitbbs.cn·[FROM: 98.]
Tuesday, May 28, 2013
发几个面经(5) Groupon 电面+onsite
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment