Loglet Lab Software Implementation

Logistic Growth and Substitution: The Mathematics of the

As of 10 July 1998, Loglet Lab for Windows 95 and Windows NT is up to version 1.1. A Tutorial for Loglet Lab is also available.

Using this version of Loglet Lab, a user can:

- fit one or more logistics, including declining logistics, to time
series data of arbitrary length using a non-linear least-squares
regression algorithm as described in section 4. For
this purpose, a ``Logistic Fitting Wizard'' can suggest initial values
for the parameters of each logistic as the point of departure for the
fitting engine;
- plot the data with its fitted component logistic(s);
- plot the residuals as described in section 3.4;
- plot the data with the Fisher-Pry transform applied, linearizing
the time-series data and its fitted logistic(s);
- determine confidence regions for each loglet parameter by using the Bootstrap
method, a powerful modern Monte-Carlo statistical technique where the user
data is used to determine the error associated with each fitted parameter.
- plot the ``bell'' view described in section 3.3, which shows
the derivative (rate of change) of a component logistic and the
discrete rate of change of the data;
- apply a logistic substitution model to multiple sets of time-series
data as described in section 5.3, including the
capacity to add hypothetical pulses as described in section
5.2;
- extrapolate a model into the past or future;
- annotate the graph (i.e., add a title and label the axes);
- print graphs using any supported Windows printer;
- export or import data to and from Microsoft Excel, or as plain text;
- consult online Help files; and
- peruse through a ``Data Gallery'' of ``classic'' sets of logistic
time-series data.

Possible improvements for future versions of Loglet Lab include:

- adding an ``Undo'' function so users can correct their mistakes.
- designing a better graphing module to provide more informative
and attractive graphs;
- alternative regression methods (robust,
);
- export a graph as a bitmap or as a Windows Metafile for use on
the World Wide Web or in presentation applications such as or
Microsoft PowerPoint;
- adding the option of constraining parameters to a range of
values, as opposed to holding to a single constant value;
- more robust algorithms for providing initial values of parameters;
- interactive visual fitting of component logistic models, where
a user could change parameters directly via a mouse. A
prototype for this can be found at
`http://phe.rockefeller.edu/applets/index.html`; - improving spreadsheet functionality (e.g., formulas); and
- improving the general user interface.

We hope users will send us suggestions to improve Loglet Lab, the
Tutorial, and this paper. Please send comments (and bug reports) to
`loglet@phe.rockefeller.edu`. We will also post
information on the status of Loglet Lab on our web site: `http://phe.rockefeller.edu/`.