EXAM II (Quantitative Business Analysis)

Descriptive Analytics

Describes/summarizes what happened in the past (summary statistics, charts, pivot tables). Lowest complexity.

Predictive Analytics

Uses past data models to predict the future or show variable relationships (linear regression, time series analysis). Medium complexity.

Prescriptive Analytics

Finds the best course of action using optimization models (portfolio finance models, price markdowns, supply network design). Highest complexity.

When to Use Tables vs. Charts:

Tables: Use when the reader needs to look up specific numerical values or make precise comparisons

Charts: Use when looking for trends, patterns, or relative comparisons.

Crosstabulation

A specialized table used to describe data involving two variables simultaneously (created via Excel PivotTables).

Scatter Chart:

Displays the relationship between two quantitative variables, often featuring a trendline.

Line Chart:

Displays quantitative data collected over time (days, months, years) with points connected to show continuity

Bar & Column Charts:

Used to compare categorical data. Bar uses horizontal bars; Column uses vertical bars.

Pie Charts:

Used to compare parts of a whole for categorical data (though bar/column charts are recommended alternatives when magnitudes vary wildly).

Bubble Chart

Visualizes three quantitative variables on a 2D graph (where bubble size represents the z-axis / 3rd variable).

Heat Map:

A 2D representation using different shades of color via Excel's Conditional Formatting to indicate data magnitude (frequently used in retail for same-store sales).

Stacked Column/Bar Charts:

Show relative values of quantitative variables stacked on top of each other within the same category

Clustered Column/Bar Charts:

Place bars side-by-side as an alternative way to compare multiple quantitative variables per category.

Time Series:

A sequence of observations on a variable measured at successive points or periods in time

Four Patterns:

Horizontal (constant mean)

Trend (long-term up/down shift)

Seasonal (recurring cycles like yearly, weekly, or daily rush hours)

Trend & Seasonal.

Moving Averages (k):

Smaller k tracks shifts faster; larger k smooths out random noise.

Exponential Smoothing (a)

Weighted average controlled by smoothing constant a. Choose a to minimize MSE; use smaller a for high random variability.

Error Metrics

actual value - forecast

Error Measures

• MAE: Average of absolute errors (avoids positive/negative cancellation).

• MSE: Average of squared errors (very sensitive to large errors/outliers).

• MAPE: Average of absolute percentage errors relative to actual values.

• Rule: Smaller error measure = more accurate forecasting method.

Simple Linear Regression

y^hat = b0 + b1x (b0 = intercept, b1 = slope)

Mathematically chooses the line that minimizes the MSE.

Goodness of Fit (R²)

Higher R² (coefficient of determination) means a better model fit.

Trend Regression

Uses time (t) as the independent variable: y^hat = b0 +b1t

Seasonality via Dummy Variables (Without Trend)

• Golden Rule: To model a categorical variable with k levels, you always need K-1 dummy variables.

• The baseline category is captured entirely by the intercept b0.

Seasonality WITH Trend:

y^hat = b0 +b1QTR1t + b2QTR2t + b3QTR3t + b4t

Qualitative Methods (Words)

Involves the use of expert judgement to develop forecasts

Quantitative Methods (Numbers)

Past information about the variable being forecasted is available.

The info can be quantified.

It’s reasonable to assume the past will continue int the future.

Average of all Past Values

uses the simple average of all historical data points available.

Naïve Method

uses the most recent data value as the forecast for the next period.

When choosing K (Moving Average)

A smaller K tracks shifts and changes in the time series MORE QUICKLY

A larger K is more effective at smoothing out random fluctuations.

using a large value of k in the moving averages method is effective in

smoothing out random fluctuations

A ________pattern exists when the data fluctuate randomly around a constant mean over

time.

horizontal

A set of observations on a variable measured at successive points in time or over

successive periods of time constitute a

time series