"Knowing the vocabulary is half the battle. Here are the essential data science terms in plain English."
Learning Objectives
By the end of this module, you will be able to:
- Recognize the most common data science terms
- Explain each in simple language
- Read data articles and docs with confidence
The Essential Terms
Bookmark this page and return whenever a word trips you up. You don't need to memorize these — that would be like memorizing a dictionary before writing a sentence. Instead, skim them now for familiarity, then come back to look up any term the moment it appears in an article, a tutorial, or a real project. Vocabulary sticks fastest when you meet a word in context and then check its meaning.
Explain like I'm new: Every field has its own shorthand, and data science is no exception. When you first hear "run some EDA on the features before we train the model," it sounds like a foreign language. Learn the handful of words below and that same sentence becomes plain English: "explore the input columns before teaching the computer." The jargon is a barrier only until you know it.
| Term | Plain-language meaning |
|---|---|
| Data science | Turning raw data into insight and decisions. |
| Data | Recorded facts — numbers, text, images, clicks. |
| Structured data | Data in neat rows and columns (tables). |
| Unstructured data | Data with no fixed format (text, images, video). |
| Dataset | A collection of data, usually a table. |
| Feature | An input column used to predict (X). |
| Label | The answer you want to predict (y). |
| Mean / Median / Mode | Average / middle / most common value. |
| Standard deviation | How spread out values are. |
| Correlation | Whether two things move together (-1 to +1). |
| Distribution | How often each value appears. |
| EDA | Exploratory Data Analysis — the first curious look. |
| Data cleaning | Fixing messy, missing, or wrong data. |
| Outlier | A value far outside the normal range. |
| Visualization | Showing data as charts to reveal patterns. |
| SQL | Language for querying relational databases. |
| NoSQL | Flexible, non-table databases. |
| pandas | Python library for working with tables. |
| Machine learning | Teaching computers to learn patterns from data. |
| Supervised learning | Learning from labeled examples. |
| Unsupervised learning | Finding structure in unlabeled data. |
| Regression | Predicting a number. |
| Classification | Predicting a category. |
| Clustering | Grouping similar data points. |
| Training / testing | Data to learn from / to check the model. |
| Overfitting | Memorizing noise; fails on new data. |
| Model | The learned pattern used to make predictions. |
| Bias | Unfair patterns learned from data. |
| Cloud | Rented computing power over the internet. |
| MLOps | Deploying and maintaining models in production. |
Key idea: Notice the flow: data (structured/unstructured) becomes a dataset of features and labels; statistics and visualization explore it; machine learning (supervised/unsupervised) models it; and ethics, the cloud, and MLOps keep it responsible and running.
Terms People Mix Up
A few of these are easy to confuse, so it's worth pinning down the differences:
- Features vs. labels: features are the inputs (what you know), labels are the output (what you want to predict). In predicting a house price, size and location are features; the price is the label.
- Correlation vs. causation: correlation means two things move together; it does not mean one causes the other (Module 5). This single distinction prevents a huge share of bad conclusions.
- Supervised vs. unsupervised: supervised learning has labeled answers to learn from; unsupervised learning finds structure with no answers given (like clustering).
- Regression vs. classification: regression predicts a number (a price, a temperature); classification predicts a category (spam or not, cat or dog).
Try this: Cover the right-hand column and read each term aloud, saying its meaning in your own words before checking. The ones you fumble are exactly the ones worth reviewing in their home modules.
Data scientist tip: Being able to explain a term in plain words matters more than reciting its textbook definition. In real work, half the job is translating jargon for colleagues who don't share it — turning "the model is overfitting" into "it memorized our old data and won't handle new cases." If you can do that translation, you truly understand the term.
Key Takeaway: These ~30 terms cover the whole journey — data types and datasets, statistics (mean/median, correlation, distribution), cleaning and EDA, visualization, SQL/NoSQL, and machine learning (features/labels, regression/classification/clustering, training/testing, overfitting), plus bias, cloud, and MLOps. Keep this as your quick reference.
Further Learning
Terms adapted from Microsoft's Data Science for Beginners (MIT License).