1 – 记录资料输入，包括数据，假設条件和参数 。
A recommended approach would contain these steps:
- Specify the input structure and data elements (including assumptions) and sources, how they are entered or uploaded/downloaded into the system (model), for example, manually, via automated processes or combination of both.
- Describe how inputs are changed and how the changes are tested.
- Provide a detailed description of the input facility for the model, such as groups, tables, variables; these should include naming conventions.
2 – 说明模型的原理和计算方式。
To effectively describe the model calculations, it is useful to:
- Describe how the model obtains its output. For instance, does the model aggregate cells from a seriatim projection? Does the model aggregate individual model cells? Does the model project a number of stochastic scenarios and determine the mean of a certain value?
- Outline what the model is projecting or representing along key dimensions (e.g., starting time, horizons, scenarios, elements that are being projected) in support of each of the model’s uses (objectives, purposes and business processes). It is important to also clarify the order of operations, such as assessment of charges and crediting of interest.
- Decide whether to use a top-down or bottom-up approach to document the calculations, using the expertise, experience and discretionary knowledge of the documenter.
3 – 记载模型如何产生结论和产生什么样的报告。
Any output structure generated by the model should be documented:
- Similar to the input, the documenter can use a data dictionary to describe the output structure.
- At a minimum, the data dictionary should contain the following two components:
- The description of line items and
- The formulaic expressions that relate them to other variables, if applicable, provided they are not documented in the calculation sections.
- Each output report may correspond to an (output) structure, as shown in Table 1.
Leave a Reply