The report delivered to the client includes, at a minimum, the statistical model and assumptions, relevant estimates, and findings, along with visual representations of the results. 

TASC reports are clear, concise, and designed to be understandable. Multiple versions of the report can be prepared if needed, depending on the intended audience or the level of detail required.

The numerical computations can be adapted into a software package for future use or integrated into other software environments. This step involves testing and monitoring to ensure everything functions correctly.

The research phase includes three key elements:

• Context

• Descriptive statistics

• Statistical model or method

The analysis begins with a thorough study of the problem’s context, the goals of the analysis, the potential applications of the results, and relevant literature and practices. 

Next, potential statistical models or methods are considered, which might include distributional assumptions, (non-)linear dependency models (e.g., regressions), segmentation, and more. 

Descriptive statistics, such as measures of central tendency, dispersion, and frequency distributions, are provided to summarize the data.

In this phase, the chosen models, assumptions, and data lead to concrete values, interpretations, and probabilistic statements. This process includes, but is not limited to:

• Estimation

• Inference

• Prediction/forecasting

Often, data pre-processing is required, which may involve extracting, merging, and rearranging data from different sources, addressing missing values, and checking for errors. 

Estimation involves calculating values for quantities of interest, inference allows for drawing general conclusions from the data, and prediction estimates unobserved data points. 

This phase often involves software-based calculations, programming in open-source environments, and packaging code to ensure replicability of results.