Business Analytics - Summary of lectures, tutorials and mock - exams
Data-driven Decision Making (DDD): the practice of basing decisions on the analysis of data,
rather than purely on intuition.
Two types of decisions
1. Decisions for which "discoveries" need to be made within data
These are problems where the objective is to unearth new knowledge or insights
from the data,
2. Decisions that repeat, especially at massive scale
In these cases, even small improvements in decision-making accuracy can have
significant impacts when applied repeatedly on a large scale.
Two major roles of data:
- data processing/engineering
- directly building models to solve business questions
Categories Big Data 1.0 Big Data 2.0
Focus Primarily on On real-time processing,
the collection and storage analytics, and deriving
of large volumes of data. insights from data.
Technologies Early big data technologies Advanced NoSQL databases,
like Hadoop, MapReduce, real-time analytics platforms,
and basic NoSQL databases. and machine learning algorithms.
Data Types Mostly structured data, although Both structured and unstruct-
some semi-structured data may ured data, including real-time
be included. data streams.
Challenges Scalability, data storage, and basic Real-time data processing,
data processing. data security, data ethics,
and advanced analytics.
Objective To handle the "Three Vs" of big data: Emphasizing real-time
Volume, Velocity, and Variety, but capabilities and actionable
with a strong emphasis on Volume. insights.
Emphasizing real-time capabilities
and actionable insights
Python’s Features
- Readable Syntax
- Extensive Standard Libraries
- Cross-Platform Compatibility
- Strong Community Support
, Python’s advantages
- Versatility and Flexibility
- Rapid Development Cycle
- Strong Ecosystem of Libraries and Frameworks
- Scalability and Portability
Workflow of business analytics
Business question -> Analytical question -> Data -> Analysis -> Insight -> Story
Four types of analysis:
1. Descriptive
a. Aims to summarize and interpret historical data to identify patterns or trends
b. It answers ‘what happened’
c. To gain insight into the underlying phenomenon or process, not necessarily to
predict
d. Example: ‘On which days of the week have video ads received the highest
viewer engagement in the past?’
2. Inferential
a. Extends beyond the available data to make inferences of the population and
based on the available example.
b. It answers the question “Is X the cause of Y?"
c. Example: ‘Based on historical data, can we infer that ads launched during
weekends generate more interaction than those on weekdays?’
3. Predictive
a. This type of analysis uses existing data to build prediction model for an
interesting outcome.
b. It answers the question "What leads to Y?” (focus is on prediction accuracy
rather than causality)
c. Example: ‘If a video ad is launched on a Friday evening, what is the predicted
engagement rate based on past data?’
4. Prescriptive
a. Forecast the future outcomes with an entry/exit of certain policies.
b. It answers the question "What would happen if...?
c. Example: ‘If we adjust the launch time of a video ad from peak hours to
non-peak hours, what impact can we expect on the viewer engagement rate?’
Techniques like regression & logistic regression can be used for both predictive and
descriptive modeling
Data-driven Decision Making (DDD): the practice of basing decisions on the analysis of data,
rather than purely on intuition.
Two types of decisions
1. Decisions for which "discoveries" need to be made within data
These are problems where the objective is to unearth new knowledge or insights
from the data,
2. Decisions that repeat, especially at massive scale
In these cases, even small improvements in decision-making accuracy can have
significant impacts when applied repeatedly on a large scale.
Two major roles of data:
- data processing/engineering
- directly building models to solve business questions
Categories Big Data 1.0 Big Data 2.0
Focus Primarily on On real-time processing,
the collection and storage analytics, and deriving
of large volumes of data. insights from data.
Technologies Early big data technologies Advanced NoSQL databases,
like Hadoop, MapReduce, real-time analytics platforms,
and basic NoSQL databases. and machine learning algorithms.
Data Types Mostly structured data, although Both structured and unstruct-
some semi-structured data may ured data, including real-time
be included. data streams.
Challenges Scalability, data storage, and basic Real-time data processing,
data processing. data security, data ethics,
and advanced analytics.
Objective To handle the "Three Vs" of big data: Emphasizing real-time
Volume, Velocity, and Variety, but capabilities and actionable
with a strong emphasis on Volume. insights.
Emphasizing real-time capabilities
and actionable insights
Python’s Features
- Readable Syntax
- Extensive Standard Libraries
- Cross-Platform Compatibility
- Strong Community Support
, Python’s advantages
- Versatility and Flexibility
- Rapid Development Cycle
- Strong Ecosystem of Libraries and Frameworks
- Scalability and Portability
Workflow of business analytics
Business question -> Analytical question -> Data -> Analysis -> Insight -> Story
Four types of analysis:
1. Descriptive
a. Aims to summarize and interpret historical data to identify patterns or trends
b. It answers ‘what happened’
c. To gain insight into the underlying phenomenon or process, not necessarily to
predict
d. Example: ‘On which days of the week have video ads received the highest
viewer engagement in the past?’
2. Inferential
a. Extends beyond the available data to make inferences of the population and
based on the available example.
b. It answers the question “Is X the cause of Y?"
c. Example: ‘Based on historical data, can we infer that ads launched during
weekends generate more interaction than those on weekdays?’
3. Predictive
a. This type of analysis uses existing data to build prediction model for an
interesting outcome.
b. It answers the question "What leads to Y?” (focus is on prediction accuracy
rather than causality)
c. Example: ‘If a video ad is launched on a Friday evening, what is the predicted
engagement rate based on past data?’
4. Prescriptive
a. Forecast the future outcomes with an entry/exit of certain policies.
b. It answers the question "What would happen if...?
c. Example: ‘If we adjust the launch time of a video ad from peak hours to
non-peak hours, what impact can we expect on the viewer engagement rate?’
Techniques like regression & logistic regression can be used for both predictive and
descriptive modeling
