微软(Microsoft)数据分析岗位面试流程

微软（Microsoft）是一家美国跨国科技企业，很多留学生在面试微软的时候也是除了很多问题，那么微软 Data Scientist的面试流程是什么样的？针对常见现场面试题目，面试者又应该如何回答？

微软数据分析岗面试流程

 

CareerInData

 

微软的数据分析面试也一共有三个流程：

 

（1）与HR的电话面试

 

第一轮面试大约持续30分钟到1小时，这轮电话面试会分为两个部分：

 

第一部分将重点考察面试者个人经历与微软公司的合适度，所以这一部分主要涉及到的问题会包括面试者的背景、简历、过往经历和项目情况等。

 

第二个部分将包括一些技术问题，这些问题主要是理论性问题，包括但不限于关于机器学习的理论问题和概率统计问题。

 

例题参考：

 

- How do lasso and ridge regression differ?

 

- Tell us how you would explain the working of a deep learning model to a business person.

 

- Explain the meaning of P-value. How differently would you explain it to customers?

 

（2）技术类筛选

 

初次筛选之后，这次面试将进行技术筛选，持续时间大概在45分钟到1小时。技术面试涵盖数据科学算法、SQL、概率统计学这三个不同的主题。

 

例题参考：

 

- You have a bag with 6 marbles. One marble is white.  You reach the bag 100 times. After taking out a marble, it is placed back in the bag. What is the probability of drawing a white marble at least once?

 

- Given a box of dimensions W, H, and coordinates of points inside that box. Find the largest area that is free of any of these points.

 

- Why neural networks work and why is it a booming field?

 

- What’s power? How to explain it to a non-statistics person? what’s a false positive and a false negative?

 

- Write a function to check whether a particular word is a palindrome or not.

 

- Find the maximum of subsequence in an integer list.  

 

- Generate a fair coin from a biased one.

 

- Generate 7 integers with equal probability from a function which returns 1/0 with probability p and (1-p).

 

- What are the ROC curve and the meaning of sensitivity, specificity, confusion matrix?

 

- Given a time series dataset, how will you predict future value?

 

- Tell us how you can compute an inverse matrix faster by playing with some computation tricks?

 

- Describe how gradient boost works.

 

- Describe the steps for data wrangling and cleaning before applying machine learning algorithms.

 

- How to deal with unbalanced binary classification?

 

- How do you detect if a new observation is an outlier? What is the bias-variance trade-off?

 

- Explain the purpose of a Support Vector Machine (SVM). Under which circumstance would you use this?

 

（3）现场面试

 

现场面试通常会持续一整天的时间，在这一轮面试中，面试者将接受5名微软数据科学家的面试。

以下五个主题是在现场面试中会单独涉及到的内容：

 

-概率统计

-数据结构和算法

-建模与机器学习系统

-租赁经理和行为面试

-数据操作

 

午休期间，面试者将于微软数据科学家进行一对一的交流，进一步了解微软以及数据科学团队所做的工作。

 

现场面试中将深入讨论前一轮技术面试中所提到的技术概念，因此应聘者要做好统计学、概率论、数据科学和机器算法的准备。现场面试可能会涉及到白板面试（Whiteboarding），应聘者可以在面试前多加了解练习通关技巧。

 

这里简单科普一下，白板面试就是让应聘者在面试时不依赖外部参考，直接在白板上手写程序。除了微软，诸如Google、Amazon等科技大厂也会采用白板面试的方法面试应聘者。

 

此外，微软的面试有很多开放式的问题，很多问题是基于数据可视化的内容，往往这些问题会有许多解决方案。

 

例题参考：

 

- Merge k (in this case k=2) arrays and sort them.

 

- Tell us the best approach to select a representative sample of search queries from 5 million?

 

- Three friends in Seattle told you it’s rainy. Each has a probability of 1/3 of lying. What’s the probability of Seattle being rainy?

 

- Explain the fundamentals of Naive Bayes? How do you set the threshold?

 

- Can you explain what MapReduce is and how it works?

 

- Can you explain SVM?

 

- How do you detect if a new observation is an outlier? What is a bias-variance trade-off?

 

- Discuss how to randomly select a sample from a product user population.

 

- How do you implement autocomplete?

 

- Describe the working of gradient boost.

 

- Find the maximum of subsequence in an integer list.

 

- What would you do to summarize a Twitter feed?

 

- Explain the steps for data wrangling and cleaning before applying machine learning algorithms.

 

- How to deal with unbalanced binary classification?

 

- How to measure the distance between data point?

 

- Define variance.

 

- What is the difference between the box plot and histogram?

 

- How to solve the L2-regularized regression problem?

 

- How to compute an inverse matrix faster by playing around with some computational tricks?

 

- How to perform a series of calculations without a calculator. Explain the logic behind the steps.

 

- Difference between good and bad Data Visualization?

 

- How do you calculate percentile? Write the code for it.

 

- Find the maximum sum subsequence from a sequence of values.

 

- What are the different regularization metrics L1 and L2?

 

- Write a code to check if a word is a palindrome.

 

所以，要想成功入职微软绝非易事。在成为微软数据科学家之前需要经过层层筛选，过五关斩六将。刚好Uoffer可以帮你在求职路上提供专业的职业发展咨询与职业技能培训服务