Quantitative Forecasting Methods Types:
There are two types of Quantitative Forecasting Methods:
 Time Series Models  are based on the premise that data patterns from the past will continue more or less unchanged into some future period.
 Causal Models  assume the variable being forecasted (dependent variable) is related to one or more other variables (independent variable) and makes forecasts based on those associations.
Time Series Methodologies or Models:
Time Series Models: a number of specific models have been developed to predict a future value (or values) from past values, Here are the most significant models, beginning with the simplest method:

Naive  Uses the immediate prior period's actual value as a forecast for the next period.
 "Next year is expected to be just like this year".

Simple Mean (Average)  Uses the average of past values as the forecast for a future period or periods.
 E.g. average the last 2 yrs as the forecast for next year.
 Simple Moving Average  Uses the average of a specific number of the most recent values as the forecast for future period(s). Values are not adjusted.

Weighted Moving Average  Uses the average of a specific number of most recent values with each receiving a different emphasis or weight.
 Average the last 3 months with the last month weight being .5, two months ago weight being .3, and three months ago weighted .2
 E.g.: Forecasted value = (1 month ago x .5) + (2 months ago x .3) + (3 month ago x .2)
 Weights must equal 1.0

Exponential Smoothing  Uses the average of a specific number of most recent values with weights assigned to each which decline exponentially as data becomes older.
 A weight is assigned to the most recent period; that weight is called "smoothing factor"
 All other weights for earlier periods are computed asa function of the smoothing factor.

Trendadjusted Exponential Smoothing  An exponential smoothing method which makes adjustments to past data when strong trend patterns are evident in the data.
 This trend can be used when the forecast period is long enough to have a trend.
 Seasonal Indexes  Adjusts past data to accomodate seasonal patterns in data
 Linear Trend Line  Uses least squares to fit a straight line to past data and extends trendline to establish forecast.
Time Series Patterns:
Over time, actual time series data will relect a pattern. Commom patterns in Time Series Data include:
 Level or Horizontal  Data are relatively constant or stable over time; little increase or decrease in values.
 Seasonal  Data reflects upward or downward swings over a short to intermediate time period, with each swing about same timing and level of change.
 Cycles  Data reflects upward and downward swings over long period of time.
 Trend  Data reflects a steady and persistent upward and downward movement over some long time period.
 Random  Data reflects unpredictable, erratic variations over time.
Time Series Decomposition:
Time series data that shows pattern often can be decomposed (i.e., separated into multiple causes).
 Decomposition is the removal of effects of each pattern component from the data.
Decomposition Process:
 1. Removing the seasonal effects from the data, typically using smoothing;
 2. Removing the overall trend from the deseasonalized data, typically using regression;
 3. Removing the cyclic effects from the remaining data values, typically using a cyclic index.
Causal Models:
Causal Models: Assume the variable being forecasted (dependent variable) is related to one or more other variables (independent variable) and makes forecasts based on those associations.
Common Causal Model Types:
 Regression Models
 InputOutput Models
 Economic Models
Causal Model Types:

Regression Models  Uses mathematical equations that relate a dependent variable to one or more independent variables that influence the dependent variable:
 Regression fits a curve to the data points to minimize error
 Curve may be linear or nonlinear
 Trend analysis uses regression with time as the independent (explanatory) variable.

InputOutput Models  Describe the flow from one stage of a process or sector to another stage or sector.
 Outputs of one stage or sector determines inputs to a subsequent stage or sector.
 Crude oil production > refining > gasoline
 Can be used at macroeconomic level

Economic Models  Specify statistical relationship believed to exist between economic quantities.
 Black Scholes model is economic model.
 Regression fits a curve to the data points to minimize error
 Curve may be linear or nonlinear
 Trend analysis uses regression with time as the independent (explanatory) variable.
 Outputs of one stage or sector determines inputs to a subsequent stage or sector.
 Crude oil production > refining > gasoline
 Can be used at macroeconomic level
 Black Scholes model is economic model.