hughec1329/res_wash

car wash BI project

goal :

references

weather

• http://casoilresource.lawr.ucdavis.edu/drupal/node/991

• http://billalex.wordpress.com/2013/04/30/r-weather/

• polled last 14 months from bom (http://www.bom.gov.au/climate/dwo/IDCJDW5002.latest.shtml) and writing script to strip out - now in 'weather.csv

• ended up using data from form at http://www.bom.gov.au/climate/data/, possible could automate through file name. in file airportweather.csv

• ftp only has current data and forecasts

dashboarding/displaying

• http://binalytics.wordpress.com/2012/06/18/quick-web-app-with-r-rook-and-ggplot2-for-a-stock-symbol-candle-chart/ using rook and ggplot
• better dynamic server - shiny
• http://www.rstudio.com/shiny/
• https://github.com/rstudio/shiny-server/wiki/Ubuntu-step-by-step-install-instructions - host locally.
• https://github.com/ramnathv/dashifyr - dashboarding framework using js plugins knob rcharts and shiny server - looks much harder to use than shinydash - need js

shinydash - more suited to future dynamic plots w data fed in??

• https://github.com/trestletech/ShinyDash - another dashboard built on shiny server and js
• nned to install gridster js library somehow?? http://gridster.net/
• awesome example, exactly what we need - http://glimmer.rstudio.com/winstontest/shiny-jsdemo/

presentation - slidify

• use html slides w inline r - like sweave but nicer to look at. can include rCharts by saving as html then loading.. look at tester.html
• slides w day, season etc

final pres

• use slidyfy w embedded r charts - e.g. 'tester'

js libraries

• rcharts - highcharts
• http://shopify.github.io/dashing/
• d3 - need to learn actual javascript of jquery

getting data

• intially export mail manually matching srearch: from -- eventually automate to scrape of entire outlook backup - pst.
• ubuntu cli readpst -S to individual plaintext files
• awk/cut/sed to strip out data from each email -- play with file 'sample_email.txt'

modelling

EDA

• ramp up time for new wash
• daily split/ weekday/weekend

dependents

• total number washes
• washes in OH
• percent premium wash
• time of washes - within oh or without

independent

• weather - can scrape from BOM, GSOD?? -- sunlight - threshold to categorical? -- precipitation - binary? -- temperature - threshold??

• weekend vs weekday

• public holiday

• season - 2 or 4?? - categorical.

• demographics - population,

• shopping centre numbers>?? can get?

findings

• day mix per wash in multibarchart
• season vs rain and sun - justify using climate data.