Table of contents
An exploratory study should always be designed to address a set of predefined questions. Our experience across numerous scientific projects has enabled us to pinpoint the five most common types of questions that researchers and data scientists typically pose.
What is an exploratory question?
Exploratory research marks the phase in the research process aimed at bridging ideas to uncover the "whys" behind potential cause-and-effect relationships. This phase is crucial when researchers begin to grasp what they are "observing" in the process of establishing cause-and-effect models.
As discussed in a previous article, exploratory questions serve a role similar to the formal hypotheses in confirmatory studies, albeit without the same level of precision and specificity.
Exploratory questions often explore the relationships between complex data groups (multidimensional), where the most relevant specific elements remain unidentified.
Five Types of Exploratory Questions
Through various scientific research projects, we have identified five types of exploratory questions:
1. Role
Focuses on understanding the influence of certain known factors on the behavior of our system or its part (responses), which are also known. For example:
What is the role that the neuronal structure has over learning and memory performance indicators of our given experimental subjects?. View related article.
2. Prediction
Aims to determine which factors will help us model certain responses in our system. For example:
Which genetic signatures will aid in predicting the evolution of tumor size in our cancer model? View related article.
3. Characterization
Seeks to identify and understand the factors that best describe our experimental groups. For example:
What are the key factors that characterize the respondents of my study by gender, age, sexual abuse history, psychological profile, and emotional condition? View related article.
4. Differentiation
Aimed at identifying what responses are the most different according to a certain already known factor. An example of such a question might be:
Which proteins are expressed differently in control group patients?
5. Thresholds
Seeks to identify the most relevant threshold values in a specific biological process. For example:
At what molecule concentration levels do we observe different dynamics in my experimental subjects?
AutoDiscovery is equipped with all necessary functionalities to address these types of questions (and more!) in an agile and effective manner. The subsequent article will detail how to apply these functionalities to each case.
Looking for Inspiration?
If you have an ongoing project or are considering developing one and want to apply automated data exploration, our inspiration tool will be very useful for framing your exploratory questions. Try it out and let us know how it goes.
Want to read more?
- Fruitful Collaborations: real success cases at applying automated EDA.
- Exploratory vs Confirmatory Research : two complementary approaches.
- Complex Data in Biomedical Research : the concept of complex data and the value of automated data exploration in biomedical research.
- How To Design an Efficient EDA Study : the 3 key ingredients of the EDA recipe.