Stop 01 · Data Science FoundationsTopics · 1/9

What Data Scientists Actually Do

In short

Movies show data scientists typing green code all night. In real life, most of the job is asking the right question and getting someone to act on your answer.

Data scientists frame ambiguous business problems as data problems, build models or analyses to solve them, and communicate results well enough that someone actually acts on them.

The mascot doctor listens with a stethoscope while holding a clipboard of test reports.

Think of it like this

Like a good family doctor

A doctor listens to your complaint, orders a few tests, reads the reports and then explains the treatment in words you understand. A data scientist does the same: understand the problem, collect data, analyse it, then explain what to do.

Watch it happen

From a vague worry to a decision

Question

Data

Analysis

Insight

Decision

1/5A manager says 'orders feel low lately'. You turn it into a real question: did weekly orders drop, and where?

The big ideas

What you need to know

Framing comes first

Turning 'sales feel off' into a clear, measurable question is often the hardest and most valuable part of the work.

Lots of data plumbing

Finding, joining and cleaning data takes a big share of your week. Models come after the data is trustworthy.

Models are one tool

Sometimes a prediction model is the answer. Often a clear chart or a simple average is enough.

Communication finishes the job

An analysis nobody understands changes nothing. You explain results in plain words, with numbers people can check.

See it in code

Walk through the code

python

1import pandas as pd2 3orders = pd.read_csv("orders.csv")4weekly = orders.groupby("week")["order_id"].count()5print(weekly.pct_change().tail(4))

1/3Load the raw orders into a table called a DataFrame.

A first look at 'are orders dropping?'. Five lines of pandas often answer the question before anyone needs a model.

Save yourself the pain

Common mistakes

Oops

Jumping straight to a fancy model the moment you get a dataset.

Do this instead

First ask what decision this will change. Then check whether a simple chart already answers it.

Oops

Thinking the job ends when the notebook runs without errors.

Do this instead

The job ends when a person understands your result and does something with it.

Your mission

Spot data science in your phone

~15 min

Quick check

A manager says 'our app feels slower this month'. What should a data scientist do first?

Go deeper

1Read 2-3 'day in the life of a data scientist' posts from different companies.
2List the last 5 product decisions a company you use was probably informed by data science.
3Notice how much of the real job is framing the problem, not just running an algorithm.