Business Analytics Notes
9/1
Analytics
- Analytics facilitates realization of business objectives through reporting data to analyze
trends
Descriptive Analytics
- Summarizing and visualizing data using graphs, tables, cubes, etc.
- Descriptive statistics - measures of centrality, dispersion, skewness, etc.
9/3
Applications
- Database marketing
- Credit risk management
- Fraud detection
- Healthcare informatics
- Email spam
- Recommender systems
- Image and handwriting recognition
CRM (Customer Relationship Management)
- Identifying profitable customers
- Better service for more valued customers
- Retaining profitable customers
- Getting a new customer costs a lot more than retaining an existing one
- A slight increase in retention could potentially reduce costs significantly
- Larger share of customer pool
Prescriptive Analytics
- Typically involves building deterministic or stochastic optimization models, and decision
modeling under risk
- Optimization is the process of selecting values of decision variables that minimize or
maximise some quantity of interest
- Steps for building an optimization model
1. Identify decision variables
2. Identify all appropriate function
3. Identify all appropriate constraints
4. Write the objective function and constraints as mathematical expressions
Summarizing Data/Descriptive Statistics
- Variables and observations
- Populations and samples
- Types of data
- Cross-sectional versus time
9/8
Populations and Samples
- Population - collection of objects upon which measurements could be taken
- Sample - subset of the population
Types of Data
- Variables are numerical and arithmetic operations can be performed on them
- Otherwise they are categorical
- Numerical variables could be further classified as interval (profit/loss) or ratio (age)
- Categorical variables could be further classified as ordinal (letter grades) or nominal
(gender)
- Variables can be discrete (# of children) or continuous (income)
9/10
Inferential Statistics
- From samples to populations
- Statistics is the process whereby data is transformed into information
- Use the information to reach conclusions about the population
Empirical Rule
- If the relative frequency distribution of measurements is approximately bell-shaped
- 68% of measurements are within 1 standard deviations
- 95% of measurements are within 2 standard deviations
- 99.7% of measurements are within 3 standard deviations
9/17
Z-Score
x−μ
- Z=
σ
- X = demand
- Mu = mean
- Sigma = standard deviation
9/29
T-Score
x−μ
- t=
s / √❑
- X =sample mean
- μ = hypothesis
- S = std deviation of sample
- N = sample size
10/1
Hypothesis Testing
Excel Functions
- =[Link](.2T) for finding T values
s
- Confidence interval based on mean - √❑
x ± t ❑α /2 ¿
- x = sample mean
- s = sample standard deviation
- n = size of sample
- =[Link]
√❑
- Confidence interval based on proportion - π^ ± Z❑α /2 ¿
- n = sample size
- π^ = sample proportion
2 2 2
- Sample size - n=z ❑α /2∗σ / E
❑ ∗σ
- E=Z α /2
√❑
Parameters
- Mean
- H0: u = u0
- Ha: u > u0, <u0, ≠ u0
x−μ❑0
- z=
σ / √❑
-
y−μ❑0
- t=
s /√❑
10/6
Hypothesis Testing
- If p is less than alpha, reject null hypothesis
- Standard error - √ ❑
π^ −π
- Critical z -
Std error
s
- Rejection region -
√❑
Chi-Squared Tests
- Test statistic is O❑i−E❑i ¿ 2
¿
O❑i−E❑i ¿ 2
¿
¿
- ¿
❑
χ 2 =∑ ¿
i
- E - expected counts, O - observed counts
10/8
Chi Squared
- Chi-squared is equal to the observed value * expected value
10/13
Assignment 2
1. Chi-squared (goodness of fit) - H0 = pi(i) = ⅛ = 18
- To
10/15
ServPro
- H0: all pay grades are equal (u1 = u2 = u3)
- Ha: At least one pay grade is different
-