A good data scientist can solve a well-defined analytical problem. A great one also tests whether it is the right problem, whether the data and success measure fit it, and how the result can inform a decision or generate better questions. These differences are not a universal scorecard: the best work depends on the role, the team and the task.
Problem definition and context
1. Starts with the request → finds the decision behind it
A request such as “build a churn model” names a task, not necessarily the need. Stronger practice asks what decision the model is meant to support, who will act on it and what would change as a result. That can reveal that the actual need is a retention policy, a better definition of churn or simply a clearer view of why customers leave. Michael Berthold describes a progression from clearly scoped model optimization toward work that formulates problems with stakeholders and, at the expert end, explores open questions.
As an Amazon Associate I earn from qualifying purchases.
2. Uses the data on hand → identifies the data needed
Available data is not automatically adequate data. Before analysis, a data scientist considers what information would answer the question, what is missing and whether it can be collected responsibly and practically. Berthold’s account of stakeholder-facing work includes identifying data to collect, rather than treating the existing table as the boundary of the problem.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
3. Treats domain knowledge as optional → works with domain experts
Statistical and computational skill cannot supply every fact about a particular setting. A clinician, operations specialist or product expert may know how a variable is recorded, what a policy change means, or which apparent pattern is a routine artifact. Data science brings statistical, computational and human perspectives together; collaboration helps make the analysis meaningful in its domain.
#1 Best Overall
- This 4-3/8" x 7" small size, 1 subject notebook has 80 double-sided college ruled sheets that fight ink bleed and are perforated for easy tear out. Perfectly sized for when you're on the go.
- Tough pockets resist tears and hold loose sheets and notes. Durable plastic water-resistant front cover helps protect your notes and our Spiral Lock wire helps prevent snags on clothes and backpacks.
- All the benefits of our larger notebooks in a smaller, easy to carry size. Sheets measure 4-3/8" x 7 when torn out.
- Available in Seaglass Green
- LASTS ALL YEAR. GUARANTEED!*
4. Accepts the metric → checks the cost of being wrong
A metric can improve while the decision gets worse. Accuracy, for example, gives equal weight to error types even when a false alarm and a missed case have different consequences. A careful analyst asks which errors matter, to whom, and whether the chosen metric reflects those costs. Berthold specifically cautions that metrics can ignore unequal costs of different errors.
5. Assumes the sample represents everyone → checks who is missing
Training data describes the people and circumstances that generated it; it does not automatically represent the people a system will face next. Berthold gives the example of relying on existing customers to recommend products to entirely new prospects: the observed group may be a poor proxy for the target group. A stronger analysis examines whose outcomes are absent or misrepresented and what that means for applying the result.
6. Treats a clean benchmark as the whole job → anticipates real-world data traps
A benchmark dataset can hide the effort and risk involved in sourcing, joining and transforming operational data. Different systems may use incompatible definitions, contain gaps or record events at different times. Data quality and provenance affect what conclusions are supportable; they are part of the analytical problem, not chores that end once a model is selected.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteRank #2
- A classroom classic: this 6-pack of 1-subject spiral notebooks helps you identify your subjects at a glance with color-coding efficiency; color assortment may vary
- The right ruling: these 8" x 10-1/2", college-ruled notebooks fit more writing per page than wide-ruled sheets; each notebook provides 70 double-sided sheets with red margin lines
- Perect perforation: Dependable micro-perforated sheets retain your must-have notes but still detach cleanly when you’re ready to revise
- Glide from page to page: Your favorite gel or ballpoint pens will move effortlessly across these smooth pages for A+ notes with minimal ink bleeding or show-through
- 3-Hold punched: Every notebook comes 3-hole punched to fit a standard binder; take along one notebook or several to save extra trips to the locker
Analytical judgment and craft
7. Reaches for a familiar algorithm → matches the method to the question
No single model family is the definition of data science. The appropriate method depends on the question, the structure and quality of the data, the consequences of error and practical constraints such as interpretability or computing resources. Data science draws on a broad set of statistical and computational methods, so method choice should follow the problem rather than habit.
8. Treats automation as an answer → recognizes where judgment is needed
Automated optimization can be useful for standard, well-specified tasks. It cannot decide on its own whether the task is framed correctly, whether the data matches the intended population or whether an unusual result has a plausible explanation. As the goal becomes less defined, creativity and domain-informed judgment matter more.
9. Optimizes a score → interprets what the score means
A headline score is not a complete account of performance. The analyst checks how it relates to the real goal, what kinds of cases it conceals and how performance changes across relevant groups or conditions. This is especially important when different errors carry different costs: a metric is useful only insofar as it supports the decision it is supposed to guide.
Rank #3
- Perfectly sized for when you're on the go, this small 2 subject notebook has 80 double-sided college ruled sheets that fight ink bleed and are perforated for easy tear out
- Tough pockets help prevent tears and hold 6" x 9-1/2" loose sheets and notes. Durable plastic water-resistant front cover helps protect your notes and our Spiral Lock wire helps prevent snags on clothes and backpacks.
- All the benefits of our larger notebooks in a smaller, easy to carry size. Sheets measure 6" x 9-1/2" when torn out.
- Made with SFI certified paper. Notebook is recyclable – just remove the reinforcement tape on the pocket and recycle the rest! Available in Blue (Color May Vary)
- LASTS ALL YEAR. GUARANTEED!*
10. Treats data cleaning as overhead → treats preparation as analysis
Joining sources, resolving inconsistent definitions, transforming fields and deciding how to handle missing values can change the result. These choices deserve scrutiny because they encode assumptions about what the data means. Berthold describes blending and transformation as substantial parts of applied work, not incidental preliminaries to modeling.
11. Looks for confirmation → investigates anomalies
An unexpected result can be an error, a data artifact or a clue that the analyst’s assumptions are incomplete. Rather than silently discarding it or forcing it into a preferred story, a careful practitioner checks its origin and asks what it could mean. In Berthold’s account of expert exploratory work, anomalies can prompt new hypotheses.
12. Runs one analysis → iterates as understanding changes
Analysis often changes the question as well as the answer. A first result may expose a data gap; a conversation with a domain expert may suggest a different comparison; a new check may weaken an initial hypothesis. Data analysis is iterative, and the next step should follow what the evidence and feedback reveal rather than a template’s predetermined endpoint.
Rank #4
- LASTS ALL YEAR. GUARANTEED! Guarantee is valid for one year from purchase or delivery date, whichever is longer. Does not cover misuse.
- Scan, study and organize your notes with the Five Star Study App. Create instant flashcards and sync your notes to Google Drive to access them anywhere from any device.
- This 5 subject notebook has 200 double-sided, college ruled sheets that fight ink bleed and are perforated for easy tear out. Sheets measure 8-1/2" x 11" when torn out.
- Tough pockets help prevent tears and hold 8-1/2" x 11" loose sheets. Durable plastic front cover is water-resistant to help protect your notes and our Spiral Lock wire helps prevent snags on clothes and backpacks.
- Made with SFI certified paper. Notebook is recyclable – just remove the reinforcement tape on the pocket and recycle the rest! Available in Pacific Blue.
13. Produces an opaque result → makes the work inspectable
A result is more useful when another person can understand how it was produced and assess whether the conclusion follows from the data and method. Clear records of inputs, transformations, code and analytical choices support reproducibility and review. Peng and Parker’s account of data science identifies reproducibility and systems engineering among the field’s important themes.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Communication and impact
14. Reports model performance → explains the decision it can support
A model score is not the outcome that most stakeholders need. They need to know what the result implies, what action it can inform and what uncertainty or limitations matter to that action. Data science is used both to inform decisions and to develop knowledge; communication is part of the analytical cycle, not decoration added after it.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Clear out junk files and repair common Windows errors3Fix the driver behind crashes, sound loss and screen glitches15. Works in isolation → builds shared understanding
Stakeholder conversations are not only a way to present finished work. They help establish the problem, surface context the data cannot show and clarify what evidence would be useful. Collaboration can also reveal that the original question needs revision. The result is a shared understanding of what the analysis can and cannot answer.
Best Value
- BEST-SELLING HARDCOVER JOURNAL: This classic 5.6" x 8" vegan leather journal features a durable and water-resistant cover, 160 college ruled lined pages, inner expandable pocket, sticker labels, ribbon bookmark & elastic closure band.
- PREMIUM PAPER: Made with high-quality, 100 gsm acid-free paper in light ivory color, our journal paper is thicker than average notebooks & note pads, so you can confidently use most pens, pencils, and markers without ghosting and bleed-through.
- LAY FLAT DESIGN FOR WRITING EASE: Our thread-bound, college ruled notebook is designed to lay flat, making it easier to write for both right and left-handed users. It’s the perfect notebook for journaling, note taking and planning.
- INNER POCKET: Includes an expandable inner storage pocket to store appointment cards, notes, receipts, and more. Personalize your journal cover & spine with the sheet of sticker labels included.
- VERSATILE LINED NOTEBOOK: Ideal for journaling, note-taking, planning, or creative writing. Whether you're making a to-do list, capturing ideas, or writing notes, this journal makes a perfect notebook for school, work, or home office.
16. Stops at a notebook or prototype → considers delivery and operation
In roles responsible for putting analytical work into use, a prototype is only one stage. The broader work may include building an application, deployment, monitoring and updates as conditions change. Whether these tasks belong to the data scientist or another team member varies by organization; the important distinction is that operational needs cannot be assumed away when a result is meant to be used in practice.
17. Measures impact by an individual model → helps the team use data better
A team’s contribution can extend beyond a single model: clearer analytical practices, better collaboration or infrastructure that lets others produce useful work. Microsoft Research’s interviews across product groups describe team leaders spreading practices as well as distinct roles contributing to actionable insight. That finding illustrates a possible form of impact, not a universal measure of an individual’s performance.
18. Assumes greatness looks the same in every role → calibrates excellence to the work
Not every data scientist should be equally strong in insight generation, specialized modeling, platform engineering, broad execution and team leadership. Microsoft Research’s 2015 report identifies five working styles—Insight Providers, Modeling Specialists, Platform Builders, Polymaths and Team Leaders—based on interviews across its product groups. It is a useful illustration of role variation, not a universal census or ranking. A fair comparison asks whether someone excels at the contribution their role requires and works effectively with complementary strengths.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →How to use these differences
These distinctions are most useful as prompts for development and team design, not as a checklist that every person must satisfy to the same degree. Berthold’s novice, apprentice and expert activity types describe different kinds of work: a clearly defined optimization task calls for reliable execution, while a poorly defined question may demand stakeholder collaboration, exploration and hypothesis generation. The progression is conceptual, not a certification scheme. For an individual, the practical question is which transition would improve the work at hand—for example, validating the metric before tuning a model, or involving a domain expert before interpreting a pattern.
For a team, complementary capability matters. One person may specialize in modeling, another in platforms or insight communication, and another in leadership. The standard is not that every analyst does everything; it is that the group can frame a worthwhile question, analyze it responsibly and make the result usable.
Sources and scope
The framing of this article is consistent with the subject of Thomas C. Redman’s 2013 Harvard Business Review article, “What Separates a Good Data Scientist from a Great One”. The 18 distinctions above are an evidence-grounded editorial synthesis, not a claim about Redman’s specific points.
Quick Recap
- Michael Berthold, “What Does It Take to Be a Good Data Scientist?” (2019), discusses novice, apprentice and expert activity types, stakeholder problem formulation, data collection, metrics, representativeness and exploratory work.
- Roger D. Peng and Hilary Parker, “Reproducible, Flexible, and High Performance Data Analysis” (2022), presents an iterative view of data analysis and highlights reproducibility and systems engineering.
- Microsoft Research, “What Do Data Scientists Do?” (2015), describes five working styles drawn from interviews across product groups.
- Benjamin F. Arnold, “Science and data science” (2018), discusses statistical, computational and human perspectives, including domain knowledge, collaboration and communication.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
