Data Analytics Services Flashcards

Question 1

What does the term data analytics mean in a professional services context?

Accepted Answer

Data analytics is the process of examining raw data to discover patterns, measure performance, and produce insights that guide decisions. In a services setting, an analytics provider collects data, cleans it, applies statistical or computational methods, and communicates findings to a client in a form that supports action.

Question 2

How does a data analytics service differ from simply buying a software tool?

Accepted Answer

A data analytics service pairs human expertise with tools, so the provider handles problem framing, data work, modeling, and interpretation for the client. Software alone only supplies functions, and the client must still know which questions to ask and how to trust the results.

Question 3

What are the four main types of analytics often used to organize services?

Accepted Answer

The four standard types are descriptive, diagnostic, predictive, and prescriptive analytics. Descriptive analytics summarizes what has happened, diagnostic analytics explains why it happened, predictive analytics estimates what is likely to happen, and prescriptive analytics recommends actions to achieve a desired outcome. Each type builds on the questions answered by the one before it.

Question 4

Why do analytics services usually begin with a clearly defined business question?

Accepted Answer

A clear business question determines which data is needed, which methods are appropriate, and how success will be measured. Without it, analysts risk producing accurate charts that do not change any decision. Scoping the question early also controls cost and timeline by preventing unnecessary data collection and rework.

Question 5

What is the difference between data, information, and insight in analytics work?

Accepted Answer

Data is a collection of raw facts or measurements, such as transaction records or survey responses. Information is data that has been organized and given context, for example a monthly sales total by region. Insight is a justified understanding of what the information means for a decision.

Question 6

Why is it important to state assumptions explicitly when presenting analytics results?

Accepted Answer

Every analysis rests on assumptions about the data, the model, and the business context, and those assumptions affect the conclusions. Stating them openly lets the client judge whether the results apply to their situation. It also makes the work reproducible, so another analyst can check the logic and rerun the analysis if conditions change.

Question 7

What role do key performance indicators play in an analytics engagement?

Accepted Answer

Key performance indicators are measurable values tied directly to goals, such as customer retention rate or average order value. They give an engagement a shared definition of success and a way to track progress over time. Choosing good indicators keeps reports focused on outcomes the business actually controls.

Question 8

What ethical considerations should guide how analytics services handle client data?

Accepted Answer

Analytics providers should collect only the data needed for the stated purpose, protect it from unauthorized access, and be transparent about how it will be used. They should respect privacy agreements and applicable legal requirements, and they should avoid presenting results in ways that mislead decision makers.

Question 9

What are the main sources from which analytics services obtain data?

Accepted Answer

Common sources include transactional systems such as sales or billing databases, operational logs from applications and devices, customer surveys, third-party datasets, and publicly available records. Each source has different strengths, such as timeliness, detail, or coverage, and different limitations, such as missing fields or sampling bias.

Question 10

What is data cleaning, and why does it take so much time in analytics projects?

Accepted Answer

Data cleaning is the process of detecting and correcting errors, duplicates, inconsistencies, and gaps in a dataset before analysis. It takes time because real-world records often arrive with mismatched formats, misspelled categories, and missing values that must be resolved or documented. Skipping this step can quietly distort every result that follows.

Question 11

What does it mean to integrate data from multiple systems into one dataset?

Accepted Answer

Integration means combining records from separate sources into a single structure that can be analyzed together. This usually requires matching shared identifiers, such as customer numbers or dates, and resolving differences in naming, units, and time zones. A well-integrated dataset lets analysts answer questions that span departments.

Question 12

Why is data quality measured along dimensions such as accuracy and completeness?

Accepted Answer

Data quality is not a single property, so analysts assess it along several dimensions that each describe a different way data can fail. Accuracy asks whether values match reality, completeness asks whether required values are present, consistency asks whether they agree across sources, and timeliness asks whether they are current enough for the decision.

Question 13

How do analysts handle missing values in a dataset?

Accepted Answer

Analysts first investigate why values are missing, since a random gap and a systematic gap call for different responses. Common options include removing incomplete records, filling values with a reasoned estimate such as a median or model-based imputation, or flagging the gap and analyzing it separately.

Question 14

What is the purpose of a data dictionary in an analytics project?

Accepted Answer

A data dictionary is a reference document that defines each field in a dataset, including its meaning, format, allowed values, source, and units. It keeps the whole team interpreting variables the same way, which prevents errors such as confusing revenue before and after discounts.

Question 15

What is feature engineering, and when do analysts use it?

Accepted Answer

Feature engineering is the creation of new variables from existing data so that a model or analysis can capture relevant patterns more clearly. Examples include computing a customer's average purchase interval, extracting the day of the week from a timestamp, or converting text categories into numeric indicators.

Question 16

How should analysts validate that a prepared dataset is fit for analysis?

Accepted Answer

Analysts check that record counts reconcile with source systems, that key fields fall within expected ranges, and that totals match known benchmarks. They also look for unexpected duplicates and confirm that transformations did what was intended. Documenting these checks gives the client evidence that the data supports the conclusions drawn from it.

Question 17

What is descriptive analytics, and what questions does it answer?

Accepted Answer

Descriptive analytics summarizes historical data to show what has happened, using measures such as totals, averages, counts, and percentages. Typical questions include how many orders were placed last quarter, which product sold most, and how customer complaints changed over time. It is the foundation for every other type of analysis.

Question 18

What is the purpose of a dashboard in descriptive reporting?

Accepted Answer

A dashboard displays key metrics and trends in one view so that users can monitor performance without reading detailed tables. Good dashboards show a small number of indicators, use clear labels and comparisons, and allow users to filter by time period, region, or product.

Question 19

What is diagnostic analytics, and how does it differ from descriptive analytics?

Accepted Answer

Diagnostic analytics investigates why an outcome occurred by examining relationships, comparing segments, and testing possible causes. Descriptive analytics reports that sales fell in one region, while diagnostic analytics asks whether the decline came from fewer customers, lower prices, a stockout, or a change in a competitor's offer.

Question 20

What is drill-down analysis, and when is it useful?

Accepted Answer

Drill-down analysis moves from a summary figure to the detailed records behind it, such as going from national sales to store-level transactions. It is useful when a total looks unusual and the analyst needs to locate the segment, time period, or product that caused the change. Drill-down is a first step in diagnosis.

Question 21

What is a cohort analysis, and why do analysts use it?

Accepted Answer

A cohort analysis groups people who share a starting characteristic, such as a first purchase month, and tracks their behavior over time. Analysts use it to separate genuine changes in behavior from effects caused by the mix of customers acquired in different periods.

Question 22

What is correlation, and why is it not proof of causation?

Accepted Answer

Correlation is a statistical measure of how strongly two variables move together, with values ranging from negative one to positive one. It does not prove causation because a third factor may drive both variables, or the relationship may run in the opposite direction than assumed. Establishing cause requires controlled experiments or careful study designs.

Question 23

How does a segmentation analysis help explain differences in performance?

Accepted Answer

Segmentation divides a population into groups that share characteristics, such as region, customer tier, or channel, and then compares performance across those groups. An overall average can hide large differences, so breaking results into segments often reveals which groups drive a trend. Good segments are meaningful and mutually exclusive.

Question 24

What does a root cause analysis involve in a diagnostic engagement?

Accepted Answer

Root cause analysis works backward from an observed problem, repeatedly asking why each contributing factor exists until an underlying driver is identified. Techniques include tracing data lineage, reviewing process steps, and testing alternative explanations against the evidence. The result should be a specific, verifiable cause and a recommendation that addresses it.

Question 25

What is predictive analytics, and what kinds of questions does it address?

Accepted Answer

Predictive analytics uses historical data and statistical or machine learning methods to estimate unknown or future values. Common questions include which customers are likely to cancel a subscription, how much demand to expect next month, and which equipment is at risk of failure. Predictions rest on past patterns, so their uncertainty should be reported.

Question 26

What is a training set and a test set, and why are both used?

Accepted Answer

A training set is the portion of historical data used to fit a model, while a test set is held back and used to evaluate how well the model performs on unseen cases. Using separate data helps reveal overfitting, which occurs when a model memorizes the training data rather than learning general patterns.

Question 27

What does a regression model estimate, and what is a classification model?

Accepted Answer

A regression model estimates a continuous numeric outcome, such as revenue or delivery time. A classification model assigns each case to a category, such as whether a transaction is likely fraudulent or whether a customer will renew. Choosing between them depends on whether the target variable is a number or a label.

Question 28

What is a time series forecast, and what components do analysts often separate?

Accepted Answer

A time series forecast projects future values of a variable measured at regular intervals, such as weekly sales or daily website visits. Analysts often separate the series into trend, seasonality, and irregular noise, because each component behaves differently and may be modeled separately. This helps avoid mistaking a seasonal peak for lasting growth.

Question 29

What is a confidence interval, and how should it be communicated?

Accepted Answer

A confidence interval is a range of plausible values for an unknown quantity, computed from sample data using a stated method and confidence level. It conveys precision, since a narrow interval suggests a more reliable estimate than a wide one. It should be reported alongside the point estimate and explained in plain language.

Question 30

What is overfitting, and how can analysts reduce its risk?

Accepted Answer

Overfitting occurs when a model fits noise and accidental details in its training data, so it performs well on that data but poorly on new cases. Analysts reduce the risk by keeping models simpler, validating on held-out data, using cross-validation, and checking whether performance drops sharply outside the training sample.

Question 31

What is prescriptive analytics, and how does it differ from predictive analytics?

Accepted Answer

Prescriptive analytics recommends specific actions by evaluating options against objectives, constraints, and predicted outcomes. Predictive analytics answers what is likely to happen, while prescriptive analytics answers what should be done about it, often using optimization, simulation, or decision rules. Prescriptive work depends on predictive results, so errors in forecasts carry forward into recommendations.

Question 32

What is optimization, and what kinds of business problems does it suit?

Accepted Answer

Optimization is the search for the best feasible choice among many alternatives, given a defined objective such as minimizing cost or maximizing profit. It suits problems with limited resources and clear rules, such as scheduling staff, routing delivery vehicles, or allocating a marketing budget across channels.

Question 33

What are the typical stages of an analytics services engagement?

Accepted Answer

A typical engagement moves through scoping the business question, discovering and preparing data, performing analysis, communicating results, and handing over or supporting implementation. Many providers add a review or validation step before delivery. Clear milestones at each stage help the client track progress and approve direction before costly work begins.

Question 34

What should a written statement of work for an analytics service include?

Accepted Answer

A statement of work should describe the business objective, the scope of analysis, the data the client will provide, deliverables and their formats, timelines, and responsibilities on both sides. It should also state assumptions, exclusions, acceptance criteria, and how changes to scope are handled. Clear terms reduce disputes and set shared expectations.

Question 35

How should analysts communicate findings to non-technical business stakeholders?

Accepted Answer

Analysts should begin with the answer to the business question, then present the supporting evidence in order of importance. They should use plain language, clear visuals with labeled axes, and explicit statements of uncertainty and limitations. Recommendations should be specific and tied to actions the audience can actually take.

Question 36

What is data governance, and why do analytics services depend on it?

Accepted Answer

Data governance is the set of policies, roles, and processes that define who may access, change, and use data, and how its quality and meaning are maintained. Analytics services depend on it because results are only as trustworthy as the data and definitions behind them.

Question 37

What is data privacy, and how do analytics providers protect personal information?

Accepted Answer

Data privacy concerns an individual's right to control how personal information is collected, used, and shared. Analytics providers protect it through measures such as minimizing fields collected, restricting access by role, encrypting stored and transmitted data, and removing or aggregating identifiers before reports are shared.

Question 38

How do analytics teams measure the business value of their services?

Accepted Answer

Teams link their work to outcomes the client already tracks, such as revenue gained, costs avoided, time saved, or decisions changed. They compare results against a baseline or a control group where possible, and they record which recommendations were adopted. Value is established through evidence of impact, not through the volume of reports produced.

Question 39

Why is it important to monitor analytics models and reports after delivery?

Accepted Answer

Data and business conditions change over time, so a model that was accurate at launch can gradually lose reliability. Monitoring tracks performance metrics, input data drift, and whether users still trust and act on the outputs. Scheduled reviews allow teams to retrain, recalibrate, or retire assets before errors affect decisions.

Question 40

What is knowledge transfer, and why do many analytics services include it?

Accepted Answer

Knowledge transfer is the deliberate handover of skills, documentation, and working methods from the service provider to the client's own staff. Many services include it so the client can maintain dashboards, rerun analyses, and extend the work without depending on the provider indefinitely. It often takes the form of training sessions, documented runbooks, and shadowing during the engagement.

Data Analytics Services Flashcards

1Foundations of Data Analytics

2Data Collection and Preparation

3Descriptive and Diagnostic Analysis

4Predictive and Prescriptive Analytics

5Delivering Services and Governance

Learning Resources