Call Girls Service Nagpur Tanvi Call 7001035870 Meet With Nagpur Escorts
Ā
Visualizations of high dimensional data using R and Shiny
1. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
Building Interactive Visualizations with
Shiny to Explore Data from Social and
Health Care in UK
Armando Vieira
2. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
Summary
ā¢ The challenge
ā¢ The inputs and the outputs
ā¢ The predictive model
ā¢ Visualizations with Shiny and Google Motion Charts
ā¢ A random walk on graphs and causality
10. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
Online demos
1. Google Motion Charts
2. Shiny + Leaflet maps
11. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
Two districts, two stories
Chiltern
ļ¼Higher Health Score
ļ¼Stable population
ļ¼Healthiest
ļ¼High satisfaction score
Liverpool
ļ¼Low Health Score
ļ¼Highest Hospital Episodes
ļ¼Economical deprivated
ļ¼High percentage unpaid social care
12. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
Conclusions I
The Health Score is:
ļ¼ Higher for less deprivated areas
ļ¼ Lower for long term illness
Not related to:
ļ¼ Health stress
ļ¼ Infant mortality rate
ļ¼ % of older population
ļ¼ Population size
13. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
Conclusions II
The Stress Score is:
ļ¼ Higher for richer districts
ļ¼ Higher for regions with large % of population > 65
Not related to:
ļ¼ % Long term illness
ļ¼ Long term disability rate
14. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
Why R?
With R and Shiny we can easily deploy interactive
visualizations dashboards for powerful data exploration
15. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
A random walk on graphs and causality
21. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
Each disease has an unique fingerprint
Lung cancer Ovary cancer
22. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
Causality?
ā¢ āMore police in precincts with higher crime; does that mean that police
cause crime?ā
ā¢ Policy decision: should we add more police to a given district?
ā¢ āLots of people die in hospitals, are hospitals bad for your health?ā
ā¢ Policy decision: should I go to hospital for treatment?
ā¢ āAdvertise more in December, sell more in December.ā But what is the
causal impact of ad spending on sales?
ā¢ Policy decision: how much should I spend on advertising?
23. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
counterfactuals, confounding variables
ā¢ āIf I go to hospital will be better off than I would
have been if I didnāt go?ā
ā¢ Sales = f(advertising) + other stuff
ā¢ Xmas a confounding variable
24. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
Beware of inferences
ā¢ The problem with doing inferences on data
originated from unknown processes is related to
the (implicit) assumption that the system and
interactions of variables are in equilibrium.
25. Armando Vieira ā Data Scientist @dataAI
Armando@dataAI.uk
How much we should be worried?
ā¢ Economics
ā¢ Experiment to determine policy change for population
ā¢ Impact of treatment on population
ā¢ Selection bias ā are random samples really random?
ā¢ Business
ā¢ Impact on advertisers who choose to use new feature or service
ā¢ Impact of treatment on those who choose to be treated
ā¢ Not necessarily worried about selection bias (but may be worried
about early adopter bias)