representations

Model Ownership and Licensing: TMA will develop licensed models at its own expense and will retain full ownership of the resulting models under the terms of a confidentiality agreement between the client and TMA.   The client may license the model by purchasing scored files generated by the model produced for the client by TMA.  Scored files will be submitted in a flat-file format and contain a record locator with the corresponding dependent variable score.

Resource Requirements: The client will provide a development file via CD-ROM, comma-separated, with labels in the header row.  A separate file in .txt format is preferred.  For each file, approximately 5,000 responders and 10,000 non-responders should be provided.  Client resources will be made available as required to obtain data, understand data codes and dictionaries, approve specification changes, and sign off on deliverables.  The client should provide a clear statement of the dependent variable, it's representation, and desired or required level of precision.  Additional categories for the same data set will require a separate model and license pricing.  It is preferred that the client provide performance metrics and baseline levels of performance from any existing models whenever possible.

Scoring Minimum: A minimum order of 500,000 records scored for a new model and 250,000 records scored for subsequent scoring runs on the same model will be required for licensed models.

Model Validation: This exercise determines whether an existing TMA model needs to be refreshed at a flat fee of $1,850.  A single file with a list of those solicited, the model scores and a flag noting who responded is needed to perform the validation.

Model Refresh: If a model's performance has decayed beyond the client's satisfaction, TMA will refresh the model with a minimum order of 360,000 records scored only if the data dictionary (i.e. file columns) is unchanged.  

Data Appends: License pricing does not include fees for third-party run-time charges and/or data appends as required for scoring.  The data append is a non-TMA service in support of significantly improved model results.  Many modeling projects require data enrichment to provide greater dimensionality, information depth and significant lift to the model.  TMA may on occasion recommend or even require a data append in order to build a licensed model.  TMA holds relationships with list vendors whereby TMA may conduct a universal data append for the purposes of building the licensed model.  The client will not incur data append costs associated with licensed model building.   However, obtaining additional data appends for variables retained by the model and required for scoring will be at the client's expense.   The appended data may be provided by the client within the file to be scored, or obtained through TMA at cost from the third-party list vendor.  TMA will notify the client prior to building a model whether a data append will be required.  TMA will then offer an unofficial estimate on behalf of the list vendor for the data append.

Independent Validation:  The client may elect to validate the model's performance.  Should this option be chosen, a separate file in addition to the resource requirements described above should be provided which contains 5,000 records.  The dependent variable should be withheld.  TMA will return the file to the client with the model scores appended for independent model validation.  The client may then match the scores against a holdout sample of responses to objectively measure performance.