Advanced Data Storytelling - Pauriccarroll/datavisualisation_udacity_q42019 GitHub Wiki

Overview

Eight Data Stories: Students will learn about sequential data stories, and how eight different story types can be used to find and tell interesting data stories.
Creating Stories in Tableau: Students will learn how to tell interactive stories in Tableau. They will learn how to add Hans Rosling Bubble Chart to Tableau and how to create a Tableau Storypoint Workbook
Project: Animate a Data Story Midterm: In this section students will apply the skills they have acquired in this advanced data storytelling course to use the World Bank Indicators data file to create an interactive Tableau Story.
Animating Visualisations: Students will use datasets for animating data and building out animations with Tableau pages and get introduced to Flourish.
Animation and Narrative: Students will learn how to add audio and narrative to their data stories using Flourish, including setting up files, charts, animations and audio files to create interactive stories.
Project: Animate a Data Story Final Project: In this lessons students will create an animated data story and add in an audio track to create a narrated flourish story they can add to their own portfolio.

Learning Objectives

Projects

Animate a Data Story Midterm Project:

Introduction: You are an analyst for The World Bank. You are conducting research for the World Bank on historical global trends to better understand the current situation and how things have developed in different countries and regions around the world. Your manager has tasked you with conducting data analysis on a world indicators data set to find and tell a data story at a conference for global development nonprofits. You are given access to historical data going back to 1990 on a variety of country-level attributes related to energy usage, the impact on the environment, and the economy.

Project Overview: In this project, you'll build on your previously developed skills of creating charts, graphs, maps and dashboards in Tableau, and apply what you've learned in the lessons on creating stories. You will use the Tableau Story Points feature and the seven data story types in order to find, create and tell a coherent data story to your audience.

Data Source here
Project Rubic here

Animate a Data Story Final Project:

Introduction: You are continuing your work as an analyst for the World Bank, understanding trends in global development by country and year. You have successfully created an interactive data story in your first project, and now your manager has asked you to take it to another level for a presentation in the auditorium.

Project Overview: Your goal of this next phase of the project is to create an animated bubble chart, make a story out of it, and add audio to narrate the story. You will be using the same dataset from the mid-term project energy_environment_economy.csv to build out your narrated story in Flourish.

Data source here.
Project Rubic here

Content

Eight Data Stories.

The following represent the eight key data stories:

Change over time: Shows how data changes over time, considering the past, the present and even potentially forecasting future trends. Examples include: Fluctuation in employment overtime, movement and swings in a particular stock valuation
Hierarchy drill down: Commonly includes hierarchical attributes - categorical elements grouped together below other elements. Examples: Business units, departments and teams. Often times we wish to compare the performance at different levels of the hierarchy.
Zoom In / Out : Start by identifying any geographic data fields like latitude/longitude, county or zip code, and create maps that show the most critical and salient measures in your data. Consider mapping circles or shapes of records, and sizing them by the number of records, or by other important fields like the amount of sales, or crime. Then, add color to these shapes associated with your most important categorical fields to see if any patterns exist. Next, try looking at the data at different levels of zoom. Start at the macro level and gradually zoom in to different zones or regions of interest. Do you see anything interesting or different? Go all the way down to the most granular level of zoom in your data. It could even be a single neighborhood or street corner. Does anything pop out at you? If so, you could be well on your way to telling a great data story with zoom!
Contrasting Values: Start with the most important measures in your data - the ones that have the biggest impact on your business, organization or community, whatever the environment. Next, rank these measures from most to least, or highest to lowest. A simple descending or ascending bar chart will do, but you might also want to consider creating a map and filtering to the most extreme values on both ends - high and low. Consider ranking individual records themselves, such as sales transactions, or you might want to aggregate them into important or meaningful categories, like product categories. Now, ask yourself what each of the items at the top and bottom of the rank ordered list have in common. Maybe all of the most profitable customers are in a certain industry, while the least profitable ones are in another. Or maybe all of the high crime areas are in a certain spot on the map, while the low crime areas are more scattered. If there’s a compelling narrative there, you might want to choose Contrasts as your dominant story type.
Intersections: This type of story, by definition, is one that involves a timeline, or perhaps an animated bar chart or scatterplot. The emphasis is on the cross-over point - when does it happen, which item or group leap-frogs over which other item or group, and why do you think this is happening? In its simplest form, the intersections data story is a timeline with lines that cross each other, or intersect, one or multiple times. Start by creating an x-axis out of your date/time variable, like sales order date, and then using a critical measure like sales or profit to create the y-axis. Use lines as your marks, and create groups by adding interesting and meaningful dimensions to detail or color. Do you see any cross-over points? If so, where, and why? Go down this road, and your intersections story type might start to take shape!
Different Factors.
Outliers: It can be fairly easy to spot outliers in a data set you’re exploring, because by definition outliers stick out like a sore thumb. It’s virtually impossible to spot them by glancing at even small tables, but there’s no easier way to uncover them than by creating charts and graphs and looking at the fringe cases. Some useful chart types for finding outliers: Simple timelines can reveal huge spikes in single variables Scatter Plots let you see how combinations of two variables can be dramatically different for specific elements Statistical charts like histograms or box-and-whisker plots can make it easy to see odd occurrences Visualize away, and ask yourself - which element or group in my data is quite different than its neighbors or peers? Examples: Athletes with abnormal abilities, celebs with phenomal wealth or influence, companies that have grown at a rapid rate.
Correlations: By definition, correlations are ways that one variable is related to another in a given data set. So to find whether you data set contains an interesting correlation, there’s no better tool than a scatter plot to show you the way. Start by creating scatterplots of your most important quantitative measures, and group the data in different ways to see if a shape emerges. Consider coloring the shapes by different categorical variables, or filtering, in order to find out if you see any interesting grouping effects in the data. Add regression lines to the points, and see whether the pattern is statistically significant (p-value), and whether the change in one variable explains the change in the other variable to a large extent or not. (R^2).Examples: Could high unemployment be related to inflations etc.

Although data visualisation often evokes comparisons to storytelling, the relationship between the tow is rarely clear articulated.`

A data story is a sequence of facts in visual form that together makes a cohesive narrative.

From a data perspective there are eight basic types of stories that are useful for convey information with stories. There stories appear again and again in newspaper articles, quarterly business reviews and articles on scientific research

EDA ( Exploratory Data Analysis): This is where you add a stage to you data analysis to determine if any of these stories are present and if so which of these stories has purpose or merit in your analysis. Is there a change over time story, a comparison story

Creating Stories in Tableau.

Overview: Creating a story in tableau, creating a new story, adding sheets to a story, Annotating story points, adding dashboards to Stories.

Key Takeaways:

Each tile in a story is pulling information from a sheet. However each tile can have distinct filters in place which will not affect subsequent tiles.
If annotation happens at the sheet level, then this annotation will follow along the rest of the story. However if we annotate on a specify part of the story this will only display it’s self there.
In the top right we have story. We also have the layout tab. This allows us to change the navigator style. Maybe we want to show numbers or dots or even arrows.
In the top right we have story. We also have the layout tab. This allows us to change the navigator style. Maybe we want to show numbers or dots or even arrows.

Animating Visualisations.

Summary of Concepts: Students will use datasets for animating data and build out animation with Tableau Pages and get introduced to Flourish.

Animating Tableau utilising the pages shelf. One of the inspirations for this course was Hans Rosling.
Note when you utilise the pages shelf and create an animation, the play button will not appear in the Tableau Public sheet.
Column is the x access, rows is the y axis. By moving year to the pages self, we can see a significant change in the parameters. We can also animate this story and make it come to life for our audience.

Comparing Flourish to Tableau

Tableau and Flourish are two very powerful data storytelling tools that have similarities and differences. They can be seen used to complement each other rather than just to replace each other, and they each have their own areas of strength.

Tableau Public allows users to connect to raw data and find interesting data stories. However, while it’s possible to create animations in the application, the “play” button does not appear in the browser after publishing*. Flourish is a data storytelling tool that’s best to use when you have formatted your data and you know what chart types to use in your data story. It has smooth animation and ability to add audio files for sharing data stories in the browser. Link to Flourish here.

Once you have created a visualisation in Flourish you will want to import data into this account. Once we have the data imported we will also want to select the right columns to visualise this data. What will represent the x axis, the y axis, how will we add information via the color, size and name. Note: To speed it up or slow it down, go to the Time Slider area of the formatting pane on the right and play with the “Time between steps when playing” control.

Animation and Narrative.

Resources

Article on Data Visualisation http://vis.stanford.edu/files/2010-Narrative-InfoVis.pdf Link to Flourish: https://app.flourish.studio/visualisation/922980/edit https://blog.revolutionanalytics.com/2017/05/tweenr.html Google Gif Maker: https://datagifmaker.withgoogle.com/ Blog Post on Google Gif Maker: https://blog.google/outreach-initiatives/google-news-initiative/make-your-own-data-gifs-our-new-too/