BUS102: Data Analytics

The four questions:

Descriptive, Diagnostic, Predictive, Prescriptive.

Descriptive:

• Historical, internal operations, reports and dashboards.

Questions: What happened? How much? How often? When? Where?

Uses: Summaries, charts, trends to understand past performances.

Diagnostic:

• Detailed past data, multiple sources. 

Questions: Why did it happen? What factors explain it? Which sectors? 

Uses: Identifies causes, patterns, root issues, comparisons.

Predictive:

•  Past+current data, external factors.

Questions: What’s likely next? Who’s at risk? When?

Uses: Forecasts, risk scores, probabilities

Prescriptive:

•  All above + rules and constraints, cost/benefit metrics.

Questions: What should we do?Which option leads to the best outcome under constraints?

Uses: Recommendations, optimized decisions, action steps

Outcome bias:

• When people simply look at the outcome to evaluate a business process.

• A cognitive bias where decisions are based on the results of past events, rather than on the processes and factors that led to those outcomes.

Ex. an investor decides to invest in real estate after learning a colleague made a big return on an investment in real estate when interest rates were at a different level.

Primary data:

• Collected first hand by businesses or organizations. 

• Internal data on customers and operations

Secondary data:

• Collected by others for a potentially different purpose but used in analysis. 

Tertiary Data:

• Combinations of primary and secondary data.

Structured data: (In terms of Big Data)

•  is data that is easily organized (or already organized) in a spreadsheet.

Unstructured data:(in terms of Big Data)

• data is more difficult to classify and needs to classified before can be analyzed

• Example: Images on Instagram, even tweets

Types of Structured data

Cross-Sectional Data, Time Series Data, Panel Data (or Longitudinal Data)

Cross-sectional data

•  Data on a sample of a population (or entire population) at a point in time.

Ex. Earnings of people in the labor force in 2021.

Time series data

• Data on a given member of a population (individual or business or geographic area) over time.

 Ex. Sales at a Café by month

Panel data (longitudinal data)

• Data on a sample of a population over time.     

• Ex. Sales at all coffee shops in Chicago over time

• The five vs. Big data:

1. Volume, Veracity (reliability of the data), variety, velocity, value (relevance of the data)

If Distribution is Symmetric:

• Mean=Median

• Either measure works fine

If Distribution is Skewed

• Mean doesn’t equal median

• The difference tells you something important!

Skewed to the Right

• mean> median

• Ex: income distribution

Skewed to the left

• Median>Mean

• Ex: Age of retirement

Variance

How far numbers are spread from the average.

• in business translations- it measures unpredictability

95% of values fall within:

2 standard deviations

• If your data follows a normal distribution you can use this rule to help identify outliers

Hypothesis testing- Testing a difference:

• Are the averages of two groups different?

• Test: t-test

• BUS ex: Do employees in the New Training Program have higher average sales than those in the Old Program?

Hypothesis Testing- Testing relationships:

• Do two variables move together?

• Test: Correlation coefficient

• BUS ex: Does increasing our advertising spend relate to an increase in monthly revenue?

r= +1

perfect positive relationship (as x increases, y increases)

r=0

No relationship

r=-1

Perfect negative relationship (as x increase, y decreases)

Correlation coefficient formula:

CORREL(array1, array2)

Interpreting P-value derived from t-tests: If the p-value is less than 0.05...

The difference between means is statistically
significant

R in excel is

the correlation coefficient, function CORREL

The 5 principles of data ethics:

• Ownership: Who owns the data?

• Transparency: Are users aware of how their data is being used?

• Privacy: Is the individual's identity protected?

• Intention: What is the purpose of collecting this data?

• Outcomes: What are the consequences of the analysis?

What is Business Analytics?

Business analytics involves using data and statistical methods to gain insights and make informed decisions. It provides a competitive advantage by revealing details about customers and operations and by helping decision-makers overcome cognitive biases.

Which Excel function gives the number of cells that meet one or more criteria?

=COUNTIFS