Please enable JavaScript.
Coggle requires JavaScript to display documents.
Presenting findings in Data Science - Coggle Diagram
Presenting findings in Data Science
Cummunicate data findings
one of the most important job skills is the ability to communicate results clearly and accessible depends on the audience
Know Your Audience
If you put too much math/statistics jargon into a report or presentation for non-technical people, they’re either going to ask you to define things for them or ask for you to clarify upon certain concepts.
Hand it over too simplified to a technical audience, however, and they’ll ask you how you came up with your results.
Structure The Sequence Of Your Story
If for example, you notice a completely unrelated trend in user behavior relative to your analysis, save it as an additional discussion point inside of an appendix at the end somewhere in your slides or report.
Find and pick relevant data
If a number of features are vague in their immediate representation and don’t explain some kind of requested result, it’s better to pass on them.
rookie mistake when trying to communicate findings through visualizations:
We shouldn’t look at a chart and think What is this chart’s purpose? The chart must be ease to understand
Choose proper methods of visualization
Some examples include bar-charts and count-plots for categorical data, time-series charts for datetime type data, and lineplots/boxplots or histograms for continuous and numeric type data.
Complexity and Purpose
Don’t let your visualizations become too noisy. You don’t need to show a plot of every single possible combination of correlations for every single variable.
Best Practices for Communicating Data Science Findings
Conclusion
Visualizations are like jokes, if you don’t immediately understand them… they’re probably not very good.
Know who your audience is.
Make sure your visualization has a purpose and is somehow tied to the ultimate conclusion of your findings because irrelevant topics sometimes make your audience lose interest.
Make sure you are using relevant data.
Choose a proper method of visualization.
Use large, short labels on your graphs so that people can quickly understand what the markers or line in a chart actually mean.
Try to simplify concepts without using any technical jargon that makes the data look messy again.
Structure of a Data Analysis Report
https://www.stat.cmu.edu/~brian/701/notes/paper-structure.pdf
Audience Analysis: Just Who Are These Guys?
https://mcmassociates.io/textbook/aud.html
https://mcmassociates.io/aud_plan_dmz.html
How to Write Data Analysis reports. Lesson -2 know your audience
https://crayondata.ai/write-data-analysis-reports-lesson-2-know-audience/
Templete
https://typeset.io/formats/search/?formatId=cd197bebdf8a10cc79f5d9556b1c4ad6
Creating a blog post
Your blog post should include the following:
A compelling title about your findings
An introduction to the data research
A section sharing the background info (definition of GDP for example) and sources for your data as well as any further research you conducted
An accompanying paragraph describing the following visualizations
A conclusion touching on the limitations of the data and further research
Visualizations that can be included:
The violin plot of the life expectancy distribution by country
The facet grid of scatter graphs mapping GDP as a function of Life Expectancy by country
The facet grid of line graphs mapping GDP by country
The facet grid of line graphs mapping Life Expectancy by country
More Resources:
The Guardian’s Datablog is a good resource for example blog posts about data visualizations.