There are many features which make CB Predictor a significant advance
over Excel's built-in forecasting capabilities.
How Excel and CB Predictor perform time-series forecasting
Using the Analysis ToolPak and other various functions in Excel,
you can perform time-series forecasting on one series of data using
one of two methods: single moving average or single exponential smoothing.
You can only try one method at a time, and you must manually set parameters
such as single exponential smoothing's damping factor or the single
moving average's number of intervals to average.
By comparison, CB Predictor lets you forecast multiple series of
historical data at once using any of eight time-series forecasting
methods, including the two methods Excel uses and additional seasonal
methods. You can forecast all the series, trying all the methods for
each one, and CB Predictor will tell you which method works best for
each (according to one of three error measures). In addition, you
can either set the parameters manually or let CB Predictor calculate
the optimal parameters for you. This means you don't need to guess
what the parameter values should be; CB Predictor will always find
the best possible parameter settings for you.
How Excel and CB Predictor perform regression
Using the Analysis ToolPak and other various functions in Excel,
you can perform multiple linear regression on one dependent variable
at a time. Excel's regression uses the method of least squares, which
can fail if the identified independent variables are a linear combination
of each other.
By comparison, CB Predictor lets you perform multiple linear regression
on one or more dependent variables with the same set of independent
variables. Also, CB Predictor's regression uses single value decomposition,
which avoids the problem with linear combinations of independent variables.
CB Predictor also has HypercastingT, which in one easy
step:
- Creates an equation that defines the mathematical relationship
between the independent variables and your dependent variable.
- Forecasts each independent variable using time-series forecasting
methods.
- Uses the equation it created in the first step, combining the
forecasted independent variable values, to create the forecast for
the dependent variable.
Output options
Excel functions place forecasted points on your spreadsheet. CB
Predictor not only places forecasted points on your spreadsheet, it
can also automatically generate reports, charts, and Excel PivotTables
from your historical and forecasted data, letting you thoroughly analyze
or present your forecasts.
Uncertainty in forecasts
By their very nature, forecasted values are uncertain. CB Predictor
is the first forecasting product to automatically create probability
distributions for forecasted values. These probability distributions
are ready to use in a Monte Carlo simulation, which can simulate the
effects of the uncertainty that is inherent to any spreadsheet model.
This feature requires Decisioneering's Crystal Ball® product.
Usability
The most obvious reason to use CB Predictor over Excel's forecasting
and regression is the CB Predictor interface. Not only does CB Predictor
let you forecast multiple series at once, compare all the methods
side by side, give you a wide range of output options, and automatically
optimize method parameters for you, CB Predictor has Intelligent Input,
which can automatically select your data series and guess other formatting
attributes. This compares to Excel, where you must jump out of the
Excel dialogs to manually select each series.
Copyright © 1999, Decisioneering, Inc. Revised 2/22/99 Technical
Note cbpred-gen-001A
