Final Flashcards
(65 cards)
1
Q
Data Deluge
A
Overwhelming flood of data being produced, surpassing the ability of organization to manage and analyze it effectively.
2
Q
What happened to data availability in recent years?
A
There has been a rapid increase in data availability, shifting from small and unavailable data to large, open datasets.
3
Q
Small Data
A
Also known as micro-data, small sets of data collected for specific objectives using surveys or consensuses. It’s manageable with basic tools, typically self-produced and not available in real-time. Used to measure phenomena and support decision making.
4
Q
Open Data
A
Publicly available data provided by governments or institutions, free and accessible. Useful for complementing company data - validates business hypothesis when other data are lacking. Not in real-time.
5
Q
Big Data
A
Complex and large datasets generated continuously and often unconsciously from various sources like social media, bankomats, ticket machines, etc.. Unstructured, informal and available in real-time.
6
Q
7V model
A
Volume
Velocity
Variety
Value
Veracity
Validity
Visualization
7
Q
Social Data
A
Data from social networking interactions. Captures user interactions, preferences and trends.
8
Q
The Information System - steps
A
It helps to transform data into knowledge
- Information Gap
- Design
- Data Collection
- Data Storage
- Data Analysis
- Information Sharing
- Decision
9
Q
3 pillars of Information System
A
- People - driving force behind data utilization.
- Technology - tools to efficiently manage and process data.
- Data - the foundation for insights and decision-making process.
10
Q
Internal Data
A
Generated and managed within an organization (ex. budgets). Should be the starting point in search for secondary data since most organizations have a lot of easy available and inexpensive in-house information.
11
Q
External Data
A
Gathered from sources outside an organization (ex. public statistics).
12
Q
Primary Data
A
Collected firsthand for a specific purpose.
13
Q
Secondary Data
A
Previously collected and processed data often used for context or supplementary insights.
14
Q
Classification of External Data
A
Published
- Government Sources
- General Business Sources
Database
- directories
- commercial
- media
Syndicated Services
- panels (home, media panel, point of sales)
- surveys (psycographic and lifestyle, general surveys, market surveys)
15
Q
Data cleaning steps
A
- Remove unnecessary information
- Filter data for relevant year
- Standardize row names and order
16
Q
Criteria for evaluating secondary data
A
- Specifications / Methodology
- Error / Accuracy
- Currency
- Objective
- Nature
- Dependability
17
Q
Types of internal data
A
- operational data
- client database
- Customer Relationship Management (CRM) and Web Analysis
18
Q
Marketing Intelligence
A
Internal and primary Data collected to understand the market environment and competitor activities.
19
Q
Ad-Hoc Surveys
A
external and primary custom-designed data collection process used when other data cannot meet specific information needs.
20
Q
Problem-identification reaserch
A
- market potential research
- market share research
- sales analysis
- forecasting
21
Q
Problem-solving research
A
- segmentation research
- product research
- pricing research
- distribution research
22
Q
Techniques used in Market Research
A
Desk Analysis
Put into context
Qualitative Research
Discover
Quantitative Research
Justify
23
Q
Structured Process in Market Research
A
- Objective definition
- Research design
- Tools preparation
- Research execution
- Analysis / Interpretation
- Presentation
- Decision
24
Q
Database Preparation
A
- Questionnaire control
- editing
- assigning codes
- processing - Data cleaning
- Analysis plan
25
Descriptive Statistics
**Count and summarize number of responses for each value of a variable using tables and graphs**. Provides a clear, condensed view of collected information. (**Frequency distribution and cross-tabulations**).
1 variable questions - How many consumers know my brand?
26
Inferential Statistics
**Assums characteristics of a population based on characteristics of a sample data**. Expands findings from small group to a larger group.
27
Relationship among data
- comparison
- deviation
- correlation
- historical series
- distribution
- share / portion
- ranking
28
Multivariate Statistics
Analyzes **relationships between multiple variables**.
_Includes **dependency**_ (**how dependent variables are influenced by independent ones**;
- _regression models_ - predicts variables based on other variables;
- _conjoint analysis_ - studies how factors affect consumer choices)
_and **interdependency**_
(**relations between variables**;
- _factorial analysis_ - examines the effect of multiple variables on one;
- _cluster analysis_ - groups similar data points to find patterns) techniques.
29
Client in Market Research
**Marketing Manager** - _initiates the projcet_ and _uses results_
**Market Research Manager** - _designs the research projcet_ and _manages the relationship with the agency_
**Top Management** - _makes final decision_
30
Supplier (agency) in Market Research
**Account** - senior researcher with commercial role
**Researcher** - designs the project
**Field Manager** - oversees data collection
**Interviewers** - conduct interviews; mostly hired externally
**Data Processors (EDP)** - analyze the data
31
Types of Data
**Numerical Data** - finite number of possible values; generated by counting; ex.: number of inhabitants
**Continuous Data** - any value within range; enerated through measurement; ex.: height of an individual
**Categorical Data** - grouped into categories; _Nominal_ - no order (ex.: gender: hair color); _Oridnal_ - with meaningful order (ex.: education level)
32
On what depends the choice of method for conducting research?
- number of data to collect
- complexity of questions
- timing constraints
- investment needed
- availability of the sample
33
Qualitative Research techniques
- in-depth personal interviews
- focus groups
- observation
- etnographic studies
- online communities
34
Quantitative Research techniques
- face-to-face interviews
- telephone interviews
- online surveys
- mail surveys
35
Building the questionnaire
- Conceptual framewor
- Writing questions
- Testing the Questionnaire
36
Types of questions
- **primary (filter)** questions - identify eligibility
- **secondary** questions - seek specific details
37
Types of answers
- open-ended
- dichotomous (yes/no)
- single choice
- multiple choice
- scales
38
Questionnaire structure
1. Filter questions
2. Classification (age, gender, etc.)
3. Core questions
4. Socio-demographic data (attitudes vs. personal care, brand image, etc.)
39
Time Series and Forecasts
Time series is an _ordered sequence of observations_ and forecasting _uses time series to predict future values based on past trends_.
40
Quantitative models in Forecasting
Quantitative methods _follow logical scheme_.
- **Extrapolative models = Time Series; focus on predicting future based on past data; "regular patterns in the past will repeat"**.
- **Explanatory models = Variable to be explained (e.g Sales)
dependent on one or more explicative variables (e.g price, adv.)**.
41
Classical approach in Forecasting
Classical tools provide systematic methods for **analyzing time series and predicting future values**.
1. Analyze the predictions problem
2. Data collection and quality check
3. Preliminary analysis
4. Model selection and estimation
5. Evaluate model fit
42
Forecast Methods
- Simple Moving Average
- Double Moving Average
- Arithmetic Mean
- Exponential Smoothing
- Decomposition (additive and multiplicative models)
43
Steps in Decomposition
Trend (T); Cycle (C); Seasonality (S); Randomness (A)
1. **Select the model** (additive or multiplicative) and **estimate the trend** (T) (C and T morfs into one variable so we don't have Cs anymore)
2. **Calculate the seasonal component** (A disappears from equation and S is pulled before the equation sign)
3. **De-seasonalization** removes seasonal effect from the time series to focus on trend and Randomness (S disappears but A comes back)
4. Using what we have, we **estimate the trend (T) more precisely** (we make linear equation T = β_0 + β_1*t)
5. **Combine estimated trend (T) and seasonal component (S) to make forecast** (Ft+m = Tt+m +/ St+m)
44
Google trends
**Google Trends analyzes and shows how often a specific term is searched on Google over time, comparing interest across regions and related topics**. The search can be adjusted to the time, place, category, a tool and related topics to what interests us.
45
Measure of Comparison between collective Phenomena
While managing a company it's important to compare data to **monitor performance and make decisions**. It involves **checking averages, comparing data over time and across different situations**.
46
What are statistical Ratios?
Statistical ratios **compare 2 related quantities to make the data easier to understand**.
- **Derivation Ratios** - Compare one group as a subset of another that influences it
- **Compositions Ratios** - How smaller part relates to the whole
- **Coexistence Ratios** - compare two related groups that doesn't influence each other
- **Density Ratios** - measure something per unit of a group size
47
Index Numbers
Index numbers **help to compare how a phenomenon changes across places and over time**.
_Temporal Index Numbers_ - compare data from different time periods.
_Spatial Index Numbers_ - compare data from different locations.
- _Simple Index Numbers_ - compares a single phenomenon to the base value on the same starting point (Fixed Base Index) or to the previous one (Moving Base Index).
- _Composite Index Numbers_ - Compare multiple Phenomena together (Laspeyres and Paasche Indexes)
48
Properties of Index Numbers
- **Identity** - If two things are the same, the index will always be 1 (or 100 if shown as a percentage).
- **Commensurability** - The index doesn’t depend on how the data is measured (e.g., kilograms or grams). Changing the unit won’t change the index.
- **Determinacy** - The index must always give a clear and specific value - there can’t be confusion about the result.
- **Proportionality** - If everything changes by the same amount (e.g., all prices double), the index will also change by that same amount.
- Time Reversibility - Switching the order of comparison (e.g., comparing 2020 to 2023 or 2023 to 2020) will give consistent results.
- Factor Reversibility - If you multiply the price index and the quantity index, the result will match the value index.
- Circularity - When comparing three points in time, the index for the first to the last should equal the product of the other two comparisons.
49
Errors in Sampling
**Sampling error** - happens when sampling doesn't match the population; higher sample = less errors.
**Non-sampling error** - mistakes during data collection:
- _coverage error_ :
Over-Coverage: Includes people who don’t belong.
Under-Coverage: Misses people who should be included.
- _measurement error_ happens due to: bad tools or methods; people not answering; wrong answers due to confusion.
50
Sample & Census
**Census** - study everyone ina group (whole population) and gives exact results;
**Sample** - study a small part if the group (part if population) to estimate the whole
51
Sampling Design Process
1. Define Objectives
2. Define Target Population
3. Define Sampling Frame
4. Select Sampling Technique
5. Determine Sample Size
6. Extract the Sample
52
How to define Target Population?
- Who you want information from (sample)
- A group that contains the sample (population)
- The location or area covered
- Time period when the study is conducted
53
Determining Sampling Frame
**A tool or list used to identify the target population.** (ex.: telephone directory; mailing lists; maps; instructions to extract data from a broader list.)
54
Errors in Sampling Frames
**Omitting Elements**: Missing some people or items that should be included.
**Adding Elements**: Including people or items that don’t belong.
55
Ways to Reduce Errors
- Adjust the population definition to match the sampling frame.
- Screen participants during data collection to ensure they belong to the target group.
- Use weighting to correct for errors in the data.
56
Probability Sampling
_**Probability Sampling** - A sampling method where each person or element has a known and equal chance of being selected._
**Simple Random Sampling** (**SRS**):
_Everyone has an equal chance of being chosen._
Pros: Easy to understand, can represent the population.
Cons: Expensive, requires a full population list.
Example: Randomly selecting 100 customers from a list of 1,000.
**Systematic Sampling**:
_Select a random starting point, then choose every i-th element._
Pros: Easier than SRS, improves efficiency.
Cons: May miss patterns if data is cyclical.
Example: Picking every 10th person on a school roster.
**Stratified Sampling**:
_Divide the population into groups (strata) and sample each group._
Pros: Ensures subgroup representation, increases accuracy.
Cons: Needs subgroup data, may miss important factors.
Example: Sampling 50 males and 50 females from a population divided by gender.
**Cluster Sampling**:
_Divide into clusters, then sample clusters randomly._
Pros: Reduces cost for large or spread-out populations.
Cons: Less representative if clusters differ greatly.
Example: Sampling 3 schools from a district, then surveying all students in those schools.
57
Non-Probability Sampling
_**Non-Probability Sampling** - Sampling methods where not all individuals have a known or equal chance of being selected. Quick and cost-effective methods, often used when precision is less critical._
**Convenience Sampling**:
_Selecting respondents who are easiest to reach._
Pros: Fast, low cost, easy to conduct.
Cons: High selection bias, not representative.
Example: Surveying shoppers at a single mall.
**Judgmental Sampling**:
_Choosing respondents based on the researcher’s judgment._
Pros: Quick, useful for specific cases or expertise.
Cons: Depends on researcher’s skill; not generalizable.
Example: Interviewing industry experts for insights.
**Snowball Sampling**:
_Existing participants recruit others into the study._
Pros: Finds hard-to-reach groups (e.g., rare traits).
Cons: Time-consuming, not random.
Example: Studying underground artists through peer referrals.
**Quota Sampling**:
_Dividing the population into groups, then sampling within quotas._
Pros: Low cost, can resemble probability sampling if bias is controlled.
Cons: Not fully representative; some key traits might be overlooked.
Example: Ensuring 40% of respondents are women and 60% are men in a survey.
58
Research Trade-offs
**Balancing Research Needs with Practical Constraints**
_Economic Trade-off:_ cost and resources vs. depth of information.
_Informative Trade-off:_ focus on key research objectives instead of perfect theoretical outcomes.
_Metric Trade-off:_ develop metrics that capture the research objectives efficiently.
59
Determining the Sampling Size
1. Importance / level of risk
2. Nature of the research
3. Number of variables
4. Nature of analysis
5. Sample sizes in similar studies
6. Occurrence of target characteristics
7. Completion rates
8. Resource constraints
- margin of error
- confidence level
- standard deviation
60
Market Segmentation Definition
**Dividing customers with similar traits into groups to better meet their demand**.
61
A Priori vs. A Posteriori Segmentation
A Priori - _uses existing company data (internal, secondary data) to divide customers by behavior but doesn't explore motivations_.
A Posteriori - _relies on custom research to identify customer needs, motivations and attitudes, creating more detailed groups_.
62
Market Segmentation Process
1. Segmentation criteria definition
2. Variables selection
3. Approach selection
4. Method selection
5. Results evaluation and segments choice
- Data Collection
- Variable Selection (choose key features that summarize consumer differences)
- Segment Creation (forms groups of similar consumers that differ from other groups)
63
Types of Variables in Data Segmentation
- **socio-demographic** (age, education, etc.) - _Easy to interpret and understand, also official statistics simplify evaluation._
- **behaviors** (spending habits, usage frequency, etc.) - _Helps to develop tailored products and marketing strategies._
- **attitudes** - _Defined by the reasons behind why consumers buy products._
- **needs and benefits** - _Functional- solve practical problems, and emotional- provide feelings like status or enjoyment. Segments created this way tend to stay stable over time, though their size and importance may change._
- **psychographic profiles and lifestyles** - _Combines psychological, sociological, demographic, and anthropological insights. Often used for media strategy and large-scale surveys._
64
Features of a Good Segment
- **Measurable**: The size and potential demand can be quantified.
- **Approachable**: The group can be reached through marketing.
- **Significant**: The demand is worth the investment.
- **Distinct**: Each segment is unique and does not overlap with others.
- **Exhaustive**: Every consumer falls into one segment.
- **Stable**: The segment remains consistent over time.
65
Market Segmentation Methods
**Factor Analysis**: Identifies patterns in data to group related variables.
- Find similar customers (market segmentation).
- Discover what features or ads people like most (product and ad research).
- Spot price-sensitive customers (pricing).
**Decision Trees**: Simple diagram that shows possible choices and their outcomes.
Companies use decision trees to:
- Find out what drives customer decisions.
- Explore how different factors (e.g., price, product features) interact.
- Identify the most important things that influence customers.
**Cluster Analysis**: Groups consumers with similar characteristics.
- Market segmentation.
- Buyer behavior analysis.
- New product development.
- Test market selection.
**Conjoint Analysis**: Measures how consumers value product attributes.
- Pick the best product features.
- Predict which product will sell best.
- Design the "perfect" product.
- Group customers by preferences.
