Tips for Final Report#
Be terse and professional.
Provide an Abstract: a very short, full summary of the research.
Present answers. Don’t present the original questions.
Don’t explain the methodology.
You may want to explain how the data is insufficient or biased.
Present the sources of the data you used, but avoid details on its structure (unless that helps understand the results). Where and how the data was collected can inform the reader of potential biases.
Be data driven and avoid speculation.
You may do some speculation, but you should have some evidence or source that you cite.
If you speculate, be sure that you’re not conflating your opinion with the facts.
In the write-up, it should be clear to the reader that you’re speculating and that you know you’re speculating.
You may want to extend beyond your peers to make your project unique and cool. You will want to find creative and valuable additions to incorporate. Here are a few considerations.
Change the research questions#
In conducting research, the formulation of a research question is a crucial step that evolves over time. As researchers delve deeper into the subject matter, gain new insights, and adapt to changing circumstances, the initial research question may undergo modifications or even a complete transformation. Changes can occur due to many reasons such as:
Pilot Study or Data Collection
Data Analysis and Initial Findings
Feedback and Peer Review
Focus on different demographics#
Having multiple research questions that focus on different areas or demographics within a project can enhance the depth and breadth of your study. Here’s a description of how to accomplish this and a specific example to illustrate the concept:
To create multiple research questions focusing on different areas or demographics, you should first identify the key dimensions they wish to explore within the broader research topic. These dimensions could be distinct variables, subgroups, or specific factors of interest. Each research question should be clear, relevant, and aligned with the unique characteristics of the subgroup or area under investigation.
Example: Let’s consider a research project examining the incidents in aerospace technology.
How does the number and rate of aviation accidents correlate with different aspects of the temporal dimension (technological progression)?
What is the relationship between aircraft mechanical factors and the likelihood of aviation accidents?
How do external environmental factors such as geographical patterns and weather conditions correlate with the occurrence of aviation accidents?
What is the role of human factors, including communication, crew-related factors, and staffing levels, in aviation accidents and incidents?
To what extent can machine learning extract factorial information from personal narratives of aviation accidents?
In this example, the researchers have formulated five distinct research questions, each focusing on a different area and demographic. By having multiple research questions, the researchers can collect data specifically tailored to each potential hazard in aerospace engineering as well as factors of ML.
The first two focus on aircraft accidents and their connection to technology. The third question covers environmental and geographical demographics. The fourth is centered towards human behavior and psychology. And lastly, the fifth questions aims to draw connections using a Machine Learning model.
The different perspectives covered in this research, enviornmental, technology, physiological, and more make it complex and incredibly insightful. It also sets up the stage for conversation between findings.
Make Connections#
Do your findings and takeaways support each other? Or contradict each other?
Once you have created your visualizations, machine learning model, or any other means of extracting information from your data, your task now is to have them communicate with each other.
Example: In the study of alcohol abuse and related factors in the United States, the following information was extracted by various datasets:
Visualizations
Regression ML Model error
Analysis on the ML Decision Tree
Feature Importance
One commonality between all of the information was that they supported that income levels can affect the rates of alcohol abuse. Highlighting these connections in your presentation can help connect fragmented parts of your research and build a deeper connection overall.