Simplifying The Concepts

Being able to generate data that supports or negates an hypothesis must start wtih developing the appropriate methodology that can help you collect, and collate the appropriate data. This is just the first part.

Besides generating the data, the second part is knowing the appropriate data analysis method, predictive method (ML), data science method etc to deploy.This must start with deploying the appropriate data processing or data preparation (pre-processing in Machine Learning) methods that will allow you check for missing values, outliers, feature or variable data types, variable recoding (called Feature engineering in ML) etc.

This must be followed with exploratory analysis method to understand what each variable or feature of the data is saying and showing. This will fundamental evolve into the next phase, which is variable or feature selection.
Feature selection involve using both statistical and machine learning methods to determine which of the variables or features is appropriate and relevant to asnwering testing the hypothesis or predictions. Depending on variable or feature type, various statistical test and machine leanining methods can be deployed. The result of the application of the feature selection method may result in variables or features reduction - a situation where features or variables that are found unimportant to the hypothesis testing or prediction are removed from the dataset.

What follows is carrying-out and deploying higher level data analysis or machine learning method that will help in predictive modelling or hypothesis testing finalization. Though post-modelling or post-hypothesis testing methods may still be deployed to chek for accuracy, reliability of the higher level models. This may depend on domain, the need to check for those reliability parameters, and the ultimate goal of the data activity.

We are the ResearchDataGuy, We simplify the Concept in the way you will understand.