Skip to content
evolve.tt

Reference

Data Work That Is Not "Become a Data Scientist"

The Evolve TeamCareersData

Most first data jobs are reporting and spreadsheets. A modelling job comes later, and many people never need one.

The IT pathways map lists data analysts, engineers, and scientists in one branch. Adverts collapse that branch into “data scientist” and attach a university-shaped fear. The work most offices actually lack is duller and more valuable: a number that is the same number every time someone asks.

  1. A spreadsheet you can explain

    One filter, one pivot, one chart with a sentence title.

  2. SQL, when copying gets old

    Ask a database for the rows.

  3. Models, maybe never

    Many good data jobs never fit one.

The order of the work

The first skill

Section titled “The first skill”

A spreadsheet you can explain.

Take a real list you are allowed to use. A club register, a stock count, a class mark sheet with names removed. Practise until you can:

  • Filter to one group.
  • Build a pivot that answers one question, such as “how many per form”.
  • Make one chart, titled as a sentence, not “Chart 1”.
  • Say which rows you deleted and why.

If the list contains other people’s private details, strip them first. The rule in what never to paste applies to a mark sheet as much as to a chatbot.

Google Sheets and Excel both count. Pick the one your target office uses. Do not buy a specialised tool yet.

What comes second

Section titled “What comes second”

SQL, when you are bored of exporting the same report. You are learning to ask a database for rows instead of copying them. A free tutorial and a sample database are enough. You do not need a warehouse, Spark, or a cloud bill.

Only after you can get the rows and explain them should you look at Python for analysis. The Python course is the right kind of start. It is not a data-science degree.

How to ignore the wrong advert

Section titled “How to ignore the wrong advert”

“Data scientist, three years of machine learning, fresh graduate” is not an entry role. How to read a job ad shows you how to separate that from “reporting analyst, Excel, we will show you the database”.

Apply to the second one. In the interview, show the spreadsheet and the sentence the chart answers. Do not show a word cloud. Do not claim a model you have not evaluated.

Many people build a good career and never fit a model. That is not a failure to become a scientist. It is the job.

In this series

Section titled “In this series”

Discussion

View all discussions on GitHub