Featured post

Textbook: Writing for Statistics and Data Science

If you are looking for my textbook Writing for Statistics and Data Science here it is for free in the Open Educational Resource Commons. Wri...

Monday 12 October 2020

Lost Chapter: Writing for your Career

 This is one of the 'lost chapters' of the textbook "Writing for Statistics and Data Science", which was removed because information changes too quickly. This chapter covers data science resumes, describing class projects to businesses, and writing letters of introduction to potential grad supervisors.

Part of this has appeared before in a 2017 post, but a few things have changed since then. For example, advised you to copy/paste a job description into your resume and to set the text size to 1 and make the words invisible, so as to get past keyword-searching bots in the resume scanning process. Gaming the system like that no longer works, so a new section on keywords is included instead.



Data Science Resumes

About sending out resumes/CVs/Applications in general.

  •  For specialist jobs like those in data science, sending a few targeted applications with cover letters is MUCH better than sending many resumes to everywhere.


  • If you’re unsure about a job, apply anyways, worry about whether it’s the best for you, or if you have a chance later. Don’t defeat yourself early.


If you meet all the requirements for a job, you're overqualified.


  • Don’t look for “statistician”, look for “data scientist”, “data analyst”, “business intelligence”, “business analyst”, “quantitative”, “predictive”.


Consider the reader:

Your audience is NOT another data scientist. It is NON-STATISTICIANS, typically. In a large company, it will be someone in a human resources department who has been told to look for certain keywords and skills. They might read your resume for 20 seconds or less.


For manual readers: HIGHLIGHT, or BOLD keywords that show up in the job posting itself. ESPECIALLY in the QUALIFICATIONS.


https://www.monster.ca/jobs/search?q=statistician   (2 entries in June 2018)

https://www.monster.ca/jobs/search/?q=Data-Scientist  (162 entries)

https://www.monster.ca/jobs/search/?q=quanitative-analyst (6 entries)

https://www.monster.ca/jobs/search/?q=data-analyst (312 entries)


Regarding keywords.


Resumes and job applications are often scanned by computer for keywords to filter out most people right away. Some, but not all, of the keywords will appear on job description itself. If you are applying cold (without being asked to by a friend in the company), then check the job posting for things that aren't 'generic' like 'data science'.

Look for synonyms of things that you do. If you talk a lot about prediction, try using the words "forecasting" or "projection" as well to increase the chance of getting by the filter with the skills you have.

Research the company you're applying to. What services do they offer? Try to search for publicly available resumes of people that might have worked in the target job before. Often the job description for hiring someone will be made based on how the previous worker described that job.

If you are lucky enough to know someone at the company, ask what words they use to describe different statistical work, and use those keywords. Don't be afraid to be direct and ask for the keywords they're looking for.

There is rarely a penalty for putting too many potential keywords. If you're trying to game the system by sneaking in an entire dictionary of keywords, that will probably be caught. Otherwise, use plenty.


In a small company, your resume may get (a little) more time and may be read by someone closer to your specialty and (slightly) more familiar with the jargon of your little corner of your field. The same guiding principle governs both cases: make it as easy as possible for someone to evaluate and say 'yes'.

What is this reader going to want to know? “How can this person fill the missing hole in my organization RIGHT NOW?”

This means that even in highly qualified personnel jobs like those of data scientists, statisticians, programmers and researchers, that the potential for long term growth within a company is not a priority, at least not at the resume-reading stage. This is a major shift from the academic world, where timelines are typically much longer.


What does this mean to you, specifically:

- Opportunities come regularly, so don't panic if you don't get the 'right' position the first time it is posted.

- Future plans like the answer to the stereotypical interview question “where do you see yourself in 5 years” are now irrelevant to many employers. They shouldn't be mentioned on resume either. Stick to what is solid: The past and present.


- Emphasize your skills as they are right now, not where they will be in 6 months. (e.g. 'I am currently studying...')   

(as a corollary, you can say you have skills or qualifications that you are almost certain to have within a month. It could take that long to start work or get an interview anyways, and by then it will already be true)


- Promising company loyalty (e.g. 'I have always wanted to work at...') in cover letters and other correspondence is a waste of time as best, and comes across as insincere at worst.  

Anybody can say they are a fan of company xxxx or product yyyy without having any qualifications. Should my dog work at a meat plant because she loves turkey?)


Anyone can present themselves as business ready with the right prep work. [1]


On the subject of transcript grades:

- After a certain minimum passing threshold, grades are not a good indicator of job performance.

- If you have graduated, or are on track to graduate soon, then you are already above this threshold, and no more about your grades needs to be said.

- The one exception would be that it is a good idea to mention the awards you have received related to grades. This includes the dean's list and scholarships. (e.g. “Graduated in 2017 with distinction”, “Made the Dean's List in 2015”)

- Hobbies and activities from high school are irrelevant unless they are programming related or you are applying to a position in fast food. In you're reading a book titled “Writing for Statisticians”, it had better be the first case or you are severely undervaluing the value of your labour.

- Rather than talk about the grades you earn or the courses you took, describe the projects you did in these courses as experiences. Be specific and clear without relying on jargon or writing too much, and keep in mind that the reader is unlikely to be familiar with the course numbers and titles from your institution.


Example 1:

Very bad: “Took Stat 485”

Bad: “Took a course in time-series”

Good:Analyzed a time-series dataset of the economy of Kansas state.”

Better:Investigated time-series econometric data, and wrote an executive report.”


Example 2:

Bad: “Took a course in big data.”

Good:Scraped, cleaned, and applied a random-forest model to police call data in a Kaggle competition.”

Also good:Developed a model to predict crime hotspots from a JSON database from the Seattle Police Department. Presented findings in a slide deck.”

In each of the 'good' examples, the experience is written in such a way as to demonstrate as many high-value skills as possible in a limited space.


The 'good' time-series example signals that

  • You (the writer) can analyze real data.

  • You are familiar with time-series data.

  • You are familiar with econometric data.


The 'better' time-series example signals that

  • You (the writer) can analyze real data.

  • You are familiar with time-series data.

  • You are familiar with econometric data.

  • You can communicate your finds to non-specialists.


The 'good' big data example communicates that.

  • You can analyze big (as in 'high volume') data.

  • You can scrape data from the web, or at least an internal database.

  • You can prepare and clean data.  (50-70% of most statistical jobs)

  • You can format results into a common government format (i.e. Kaggle).


The 'also good' big data example communicates that.

  • You can analyze big (as in 'high volume') data.

  • You can build predictive, actionable models.

  • You can work with JSON data.

  • You can disseminate your findings to non-specialists, such as experts in fields other than your own.


Use 'business language' to subtly stretch the truth and frame things more favourably. For example, use the word 'setback' instead 'of failure', or use 'leverage' instead of 'use' or 'exploit'.


Use 'action verbs' as a helpful guide to demonstrate your experience, especially in the first work of each statement of your experience. These are verbs that typically imply leadership, teamwork, or productivity skills. Such words include, but are not limited to:


(distributed, produced, created, developed, disseminated (i.e. spread), maintained, updated, cleaned, scraped, prepared, built, wrote, analyzed, coded, investigated)



In your experience and even your education, try to start as many sentences as possible with one of those action words. Remember to write about what you DID and not what you DO. In other words, use the past tense for everything including your current position.

One apparent exception is when describing duties instead of actions. A popular way to write about duties instead of actions is to describe duties is to write “responsible for...”. This sounds like it's present tense, however, it's short for the past tense “I was responsible for..”, which brings us cleanly to the next point:


Taking advantage of assumptions and formatting:

You may have noticed some things are missing from the 'good' examples of experience. Specifically, articles and some prepositions are missing. The statements on a resume should be closer to news headlines than to complete sentences.


Everything on a resume is assumed to be about the person whose name is at the top of the resume. “Statements that start with 'I was' are already longer than necessary; you and the things you have done are the topics of your resume, so it makes the most to include key information only, as long as it is not ambiguous. The other relevant 'what's and 'who's in a good resume are typically made clear from formatting.


Consider the following example:

“Constructed the database management system for the company”

 which can be shortened to

“Constructed database management system.”

while retaining all or nearly all of the meaning in a resume standpoint. In this example, the article “the” isn't necessary because without the tests. Likewise, “for the company” is redundant.


Who else would you be doing this work for, if not the company? (If it was a personal skill building exercise, you would still leave that information out, the point is that you have the skill. Why you got it is not important.)

The fewer words you use, while retaining the meaning, the less of those precious 20 seconds of reading time will be to ensure as great as possible a share of that time is spent observing that you have the qualifications requested in the document.


Additional resources:






Examples: Data Science Resumes


The following excerpts come from a graduate student's resume before a revision into a more business friendly format. Take these passages and rewrite them according the principles of this section.


1. “This thesis develop a diagnostic tool for checking continuous distribution’s normality, because it is an important assumption to use Integrated Nested Laplace approximation(INLA)”


Possible solution:

1) Developed a diagnostic tool for checking normality in order to use Integrated Nested Laplace approximation (INLA) as an replacement to using Markov chain Monte Carlo (MCMC).



2. Two main projects are finished. The first one is about customer behaviours that how to distinguish a online buyer and a offline buyer, this project is accomplished by using Hidden Markov Model(HMM). The second project is about push notification effect, this project is accomplished by three steps. The push notification effect is defined as click through rate(CTR), it is estimated by two different approaches, and the best model is chosen by out of sample test.


Possible solution:

Modeled customer behaviours that how to distinguish online buyers and offline buyers using Hidden Markov Model (HMM) on click stream data. Investigated the push notification effect on different conversion metrics.


3. Detailed achievements:

Understand the company’s database.

Discover for right question.

Solve questions in a quantitative way:

Define the question

Data clean

* Deal with outliers

* Check data source



How to present analyzed result to different audiences.

How to face failures.


Possible solution:


Other activities:

Mastery of corporate databases.

Framing of research questions.

Quantitative Problem solving

Data cleaning and verification

Modeling and inference

Presentation of analysis results to different audiences.

Perseverance in the face of setbacks and difficulty.



4. My main job is to assistant Postdoctor [name], Associate Research Fellow [name], and Research Associate [name] to do extensional research on [research or grant name]. At the same time, I assist the students in the study group on choosing statistic methods.

Possible solution:

Assisted Postdoctor [name], Associate Research Fellow Dong Xu, and Research Associate [name] to do extensional research on ''Application of constructed wetland for water pollution control in China during 1990–2010''. Assisted students in choosing statistical methods.



5. The TA is response for helping the student in first and second year statistical courses, and marked the midterms and final exams of these courses. In addition, I marked the assignments for two statistical courses

Possible solution:

Face to face assistance to students in first- and second-year statistical courses, grading of assignments and exams.



6. This case competition needs to solve five questions by using the car sensor data. Our team(3 people) aims to finish three questions, and I finish the question that distinguish that a car is driving in city or highway. Here is the showcase Shiny App (link).

Possible solution:

Worked in a team to answer questions using car sensor data. My primary contribution was distinguishing when a car was driving in city or on a highway. Created an interactive showcase of results as a Shiny App (Click to see)



7. This project is about to design the best way to travel the breweries in the lower mainland. The approach is to optimize a function which consider the traveling distance and the rate of beers.

Possible solution:

Designed a method to find the optimal way to travel the breweries in the lower mainland based on traveling distance and customer ratings of beers.

Letters of Introduction (e.g. to grad school supervisors)


This section is about making a cold contact through a letter of introduction. That is, an introduction of yourself without being prompted or solicited to from someone else. Being able to do this will greatly improve your networking prospects.


Start with a 'My name is... I am a [job / student]” paragraph.


For clarity use the phrasing 'My name is [name]' and not 'I am [name]'.


Immediately after the 'my name is' paragraph, make it clear why you are writing the letter and what you want.


One common goal of a cold introduction is to get an 'informational interview'. In such an interview, the goal is not to see if you're a good fit for a certain job position, but with the company, school department, or organization as a whole. This is your opportunity to learn things about the organization that wouldn't be easy to find by searching online. See the footnote at the end of this section.


A letter of introduction is NOT a cover letter. A cover letter is specific to a job or position, whereas this is much more general. A letter of introduction is not a reference letter either, for similar reasons.


Your letter of introduction should demonstrate that

  •     You’ve done the research to see what this prof’s work is about.

  •     You have a specific goal.

  •     You’ve put time into this introduction (instead of just mass-sending many similar introductions).


Letter of introduction template.

Dear [person, be sure to use the proper honorific (e.g. Mrs, Mr, Dr). If you are unsure if a professor has a PhD or not, use 'professor']

My name is [your name], and I am currently [your defining position right now, such as a 4th year undergrad].


I am interested in [your goal, such as doing a master's degree or getting a position in financial analysis]*. [Other thing that you did that led you to have this goal]. [I am interested in doing what you're doing or similar (job), or interested in doing research in your area, etc.]**. [Mention of plan for this relationship]*** I would appreciate hearing from you about this line of work / research (prompting the contact to talk about themselves or their own work, which people typically enjoy doing)


I have included my resume / CV for review (note that you are not assuming this is for a particular position, but just in case the contact would like to have it on hand, or know more about you). If you have time to meet with me, [ways that I can be contacted].

Thank you for your time!

[Your name again]


* and ** These two parts are the most important.


**If you are trying make a contact with a particular company, you would talk about some previous work experience that was similar or comparable to what you are trying to get into.


*If you are trying to get into graduate school, this is where you would mention what you want to study in your program. Be as specific as possible, but don't try to use terms or jargon that you don't understand – using fancy terms won't impress anyone. A good example would be “I want to attend graduate school in applied statistics with a focus on neural networks.”


**If you are trying to get into graduate school, this is where you would mention a previous course or project that you did that led you to this contact as potential supervisor. Here you can be VERY SPECIFIC. If it was a class, give the name of the class and the name of the professor that taught it (there's a chance that your potential supervisor knows of the professor). If it was an article or book, mention the title, ESPECIALLY if it was something the potential supervisor wrote.


*** For a company, this would be when you would ask for an informational interview with “I would appreciate hearing from you about this line of work / research”. This is an invitation to your contact to talk about themselves or their own work, which people typically enjoy doing.


*** For a potential supervisor, here is where you would VERY briefly describe the kind of research project you intend to you in your program. This is NOT a commitment to work on that project, but it does show that you have at least thought about the research you would be doing and have made a plan.


For more details for industry, see https://www.thebalance.com/letter-of-introduction-examples-and-writing-tips-2062593 By Alison Doyle

For more details for academia, see https://theprofessorisin.com/2011/07/25/how-to-write-an-email-to-a-potential-ph-d-advisor/ By Dr. Karen Kelsky

For further interest, see: https://www.mcgill.ca/gradapplicants/research-supervision/connecting-supervisor


Footnote: As a personal example, in an informational interview with a current graduate student when I was applying to graduate school, I found that the worst thing about the department I was applying to was the rainy weather in winter. In other words, there were no academic issues with the department worth mentioning. Without talking to someone in the department directly, I would have had to rely of the word of people speaking in a public forum, where complaints could be discouraged. In private interviews, people tend to be a lot more candid and honest.

"My Minion Little Boo" by DaPuglet is licensed with CC BY-SA 2.0. To view a copy of this license, visit https://creativecommons.org/licenses/by-sa/2.0/

No comments:

Post a Comment