What is Data Analysis, Really?

When people hear the phrase data analysis, many immediately picture spreadsheets full of numbers, colourful charts, and complicated formulas. Others imagine machine learning models or artificial intelligence discovering hidden patterns.

Those tools certainly have their place. But they can also obscure a more fundamental truth:

Data analysis is not about calculations. It is about answering questions.

The calculations matter only because they help us answer those questions.

For a textsmith, this perspective feels surprisingly familiar. Every day we search, filter, transform, compare, and summarize text. Those are analytical activities. The only difference is that sometimes our “data” consists of words instead of numbers.

Perhaps it is time to rethink what data analysis really means.

Data Is Not Information

One of the most important distinctions in statistics is the difference between data and information.

Data are raw observations.

A weather station records today’s temperature.

A shop records every sale.

A web server records every visitor.

A shell script records the output of a command.

These are all pieces of data.

Information emerges only after we understand what those observations mean.

Suppose someone tells you:

Yesterday we sold 3,452 products.

That is still mostly data.

Now suppose they tell you:

Yesterday’s sales were 28% higher than our average Friday because a marketing campaign launched that morning.

Now we have information.

The numbers have acquired context.

Analysis is the bridge that transforms data into information.

Questions Come Before Calculations

One mistake beginners often make is opening a spreadsheet before deciding what they actually want to know.

The experienced analyst does the opposite.

They begin with questions.

• Which products sell the most?
• Which customers return most often?
• Which articles attract readers?
• Which commands fail most frequently?
• Which expenses are increasing fastest?

Only after defining the question do they decide which calculations are necessary.

Without a question, there is no analysis—only arithmetic.

This principle extends well beyond statistics.

When using grep, we first ask, “Which files contain this pattern?”

When using find, we ask, “Where are these files located?”

When using sort, we ask, “What happens if these records are ordered?”

The tool never comes first.

The question does.

Collecting Data Is Not Analysing It

Modern computers make it incredibly easy to collect enormous amounts of data.

Websites log every visitor.

Phones record locations.

Applications generate diagnostic logs.

Businesses store years of transaction histories.

Sensors produce measurements every second.

Storage has become cheap.

Meaning has not.

Many organisations proudly announce they have collected terabytes—or even petabytes—of data.

That alone tells us very little.

Imagine someone who spends years collecting books but never reads them.

We would not call them a scholar.

Likewise, simply collecting data does not make an organisation data-driven.

Analysis begins only when someone starts asking meaningful questions about those records.

Data Analysis Is Pattern Recognition

Every analytical task, no matter how sophisticated, boils down to looking for patterns.

Sometimes we seek similarities.

Sometimes differences.

Sometimes trends over time.

Sometimes unusual events.

Sometimes relationships between variables.

Even the simplest Unix commands embody this idea.

grep searches for matching patterns.

uniq discovers repeated values.

sort reveals order.

diff highlights change.

wc summarizes.

These are analytical tools because they help us identify patterns hidden inside data.

Statistics simply extends these ideas using mathematical methods.

Why Textsmiths Should Care

Textsmiths often assume that data analysis belongs to statisticians.

In reality, we perform analytical work constantly.

Suppose you receive thousands of log files.

You search for failed logins.

You count their frequency.

You group them by machine.

You identify the busiest hours.

You produce a summary report.

That entire workflow is data analysis.

The data happened to be text.

Likewise, analysing Markdown documents, source code, CSV files, JSON data, email archives, configuration files, or web logs all involve the same intellectual process:

collect, inspect, transform, summarize, and explain.

The distinction between “text processing” and “data analysis” is therefore much smaller than many people imagine.

Beyond Numbers

Statistics textbooks often define data as numerical observations.

Modern computing has broadened that definition considerably.

Today we routinely analyse:

• text
• images
• audio
• video
• source code
• network traffic
• system logs
• genomic sequences

Not all of these are numeric.

Yet they are all data.

The analytical mindset remains exactly the same:

Ask good questions.

Find relevant evidence.

Transform the raw observations.

Communicate the conclusions.

Analysis Is Thinking

Perhaps this explains why learning spreadsheets alone rarely makes someone a good analyst.

Nor does learning Python.

Nor R.

Nor SQL.

Nor machine learning.

Nor artificial intelligence.

Those are valuable tools.

But they cannot replace clear thinking.

The best analysts spend more time refining the question than writing formulas.

The calculations are often the easiest part.

Understanding what problem deserves solving is the difficult—and rewarding—work.

The Textsmith’s Perspective

At The Textsmith, we often say that computing is fundamentally about processing text.

Data analysis fits naturally within that philosophy.

Whether we use shell pipelines, SQL queries, Python scripts, R notebooks, or statistical software, we are following the same journey:

Raw observations become meaningful information.

The medium changes.

The thinking does not.

Perhaps that is the simplest definition of data analysis.

It is not the art of producing numbers.

It is the discipline of turning raw data into understanding.

And that is something every textsmith already knows how to begin doing.

• Alternative Ways

• Data Science
• Markdown
• Spreadsheets