presentation-notes.md

• Google slides
• Connecting to a file
  • Several different file types
    • JSON
    • PDF tables
    • Google sheets
    • Dropbox
    • With the paid version - mySQL, PostgreSQL, salesforce
  • Click Excel file - navigate to FBI file
• Cleaning up the data
  • Load in excel sheet of FBI crime data
  • Use Data Interpreter - Review the results
  • 2nd tab - point out new headers, merged cell handling
  • Merge sexual assault columns
  • Other neat features - pivoting (spreadsheet where each year was a column and the value is a population), splitting text into multiple columns
• Tour of sheet 1 tab
  • Dimensions - categorical
  • Measures - numerical / anything it makes sense to sum or take an average of
  • Marks card - Altering the appearance of the graph
  • Show me - basic graph types to choose from
• Demonstration #1
• First graph - sum of robberies by state
  • Note that you can change the sum to a median or an average
  • Sort results
  • Show mark labels
  • Color - can change it manually OR command drag Robbery to color
    • Change the color scale to a different scale
  • Click the plus sign to see cities makeup
  • Let’s add to hierarchy
  • Right click on population - create - bins
  • Drag population (bin) to between state & city in hiearchy
  • Call the sheet - Robberies
• Save the workbook
  • Create a Tableau Public account if you haven’t yet
  • Log into account
  • Check in about pacing
• Violent crimes by cities map
  • Start with counts of violent crimes
  • Filter on state Colorado
  • Violent crime rate = violent crime/population * 100000
    • A crime rate describes the number of crimes reported to law enforcement agencies per 100,000 total population.
  • Replace violent crime
  • Examine data under Lakeside, then exclude
  • Drag the size up
  • Label unknowns
    • 39.6011 -105.0322
    • 38.9122 -106.9624
  • Call the sheet - Violent Crime Rates
• Arson by burglary
  • Rename arson variable
  • Right click trend lines - show trend lines
  • Hover over line - see p value is very small
  • Drag state onto color to add more detail - drag it off, since too many measures
  • Name the sheet - Arson vs Burglary
• Making a dashboard
  • Change size to automatic
  • Drag in all graphs
  • Use Robberies by state as a filter
  • Add to tooltip - population for map
  • Name sheet - Analyzing 2013 FBI Crime Data
• Breakout #1
  • Average arson by state
  • Sum population on color
• Demonstration #2
• Titanic data
  • Text file import
• Cabin group survivors
  • Move survived & pclass to dimensions
  • Make aliases for survival & class
    • Just for my reference: port of embarkation (C = Cherbourg, Q = Queenstown, S = Southampton)
  • Cabin group = LEFT([Cabin], 1)
  • Cabin group & survived as the left side of the graph
  • Make a group from Unknown & T
  • Table calculation -> Percent of total -> specific dimensions -> check Survived
  • Survived onto color
  • Percent onto label
  • Number onto label
  • Drag to see graph better
• With family
  • With family calc
    • IF [Parch] = 0 THEN 'No Family'
    • ELSEIF [SibSp] = 0 THEN 'No Family'
    • ELSE 'With Family'
    • END
  • Bubble graph
  • Change color palette just for fun
  • Add number for label
• Ages and Classes - Box & Whisker
  • Age by sex & class
    • Sex on top, class on bottom — age as the DV
  • Colored by class
  • Change age default to no decimals
  • Average Fare in _
    • Drag fare to tooltip
      • Look at the underlying data - is just giving us individual fares for each datapoint
    • {FIXED [Pclass]: AVG([Fare])}
  • Change fare format to currency
  • Make dynamic tooltip
• Titanic dashboard
  • Resize traveling with family graph to fit automatically
  • Add filter for traveling with family dropdown - apply to all
  • Add filter for age - apply to all
  • Add little image of titanic
• Breakout 2
  • {FIXED [Ticket]: COUNT(Name)}
  • Age vs number of people on ticket
  • Trend line
  • Add to dashboard
  • Add filter
• Demonstration #3
• Billionaire dataset
  • Add currency as default
• Gender statistics
  • Parameter - Gender Graphs
    • Average Age
    • Count of Billionaires
  • Dynamic Measure
    • IF [Gender Graphs] = 'Average Age' THEN AVG([Age])
    • ELSEIF [Gender Graphs] = 'Count of Billionaires' THEN SUM([Number of Records])
    • END
• Top Billionaires
  • First show top 10 fixed billionaires, then show adding parameter functionality
  • Top Billionaires parameter - range 1-50
  • Filter name by worth in billions and top billionaires parameter
  • Circle graph
• Tabby
  • Need to install
  • Activate from the command line
  • mdfind kind:folder "tabpy_server"
    • sh /Users/applemacbook/anaconda/envs/Tableau-Python-Server/lib/python2.7/site-packages/tabpy_server/startup.sh
    • Recommend making an alias in your bash profile if interested
  • Help -> Settings -> Manage External Service Connection
    • Port 9004 by default
    • Local host
  • SCRIPT_BOOL
  • SCRIPT_INT
  • SCRIPT_REAL (like float)
  • SCRIPT_STR
  • Use _arg# to reference Tableau measures or dimensions
• Countries by worth
  • Country Name
    • SCRIPT_STR("
    • return [x.title() for x in _arg1]
    • ",
    • MIN([Country of Citizenship]))
  • Average Worth
    • SCRIPT_INT("
    • return [x*1000000000 for x in _arg1]
    • ",
    • AVG([Num Worth in Billions]))
• Billionaire Analysis - Story
  • Make a story - guided dashboard/findings
  • Resize graphs at source
  • Billionaire couples tend to be older than single billionaires.
  • Most of the billionaires, however, were men.
  • In fact, in the top 10 billionaires, none of them were women.
  • Interestingly, billionaires in Mexico, Nigeria, and Saudi Arabia had the highest average wealth.
  • Here's my next steps.
    • For my next project, I'm going to focus on successful women and what made them successful!
• Breakout 3 - choose your own adventure!
• Feedback survey!