ETC1010: Data Modelling and ComputingLecture 9B: More on networksDr. Nicholas Tierney & Professor Di CookEBS, Monash U.2019-10-091 / 38

AnnouncementsAssignment 3 is due this Friday at 5pm
Project deadlines:Deadline 3 (11th October) :  Electronic copy of your data, and a page of data description, and cleaning done, or needing to be done.
Deadline 4 (18th October) :  Final version of story board uploaded. 

Guest Lecture: 16th October: Dr. James McKeone 
Practical exam: 18th October in class at 8am
Final Exam: I will provide a review of exam content

2 / 38

recap: on network dataTo make a network analysis, you need:an association matrix, that describes how nodes (vertices) are connected to each other
a layout algorithm to place the nodes optimally so that the fewest edges cross, or that the nodes that are most closely associated are near to each other.


3 / 38

Quantitative association matrices

Previous association matrices were black and white:

4 / 38

Quantitative association matricesYou could have the association between nodes described as real numbers.

5 / 38

Quantitative association matricesE.g., these are the number of times that these people called each other in the last week:


 
      
    Meg 
    Tay 
    Yat 
    Zili 
    Jess 
  


    Meg 





  

    Tay 





  

    Yat 





  

    Zili 





  

    Jess 





  

/ 38

	Meg	Tay	Yat	Zili	Jess
Meg	0	5	4	1	1
Tay	5	0	4	2	1
Yat	4	4	0	0	0
Zili	1	2	0	0	6
Jess	1	1	0	6	0

Quantitative association matrices

We would need to turn this into an edge data set:

## # A tibble: 25 x 3
##    from  to    count
##    <chr> <chr> <dbl>
##  1 Meg   Meg       0
##  2 Tay   Meg       5
##  3 Yat   Meg       4
##  4 Zili  Meg       1
##  5 Jess  Meg       1
##  6 Meg   Tay       5
##  7 Tay   Tay       0
##  8 Yat   Tay       4
##  9 Zili  Tay       2
## 10 Jess  Tay       1
## # … with 15 more rows

7 / 38

Quantitative association matrices

We need to decide what corresponds to a "connection".
Let's say they need to have called each other at least 4 times, to be considered connected.

d_edges_filter <- d_edges %>% filter(count > 3)

8 / 38

Quantitative association matrices

d_edges_filter

## # A tibble: 8 x 3
##   from  to    count
##   <chr> <chr> <dbl>
## 1 Tay   Meg       5
## 2 Yat   Meg       4
## 3 Meg   Tay       5
## 4 Yat   Tay       4
## 5 Meg   Yat       4
## 6 Tay   Yat       4
## 7 Jess  Zili      6
## 8 Zili  Jess      6

9 / 38

Quantitative association matrices: Make the network diagram.

library(geomnet)
set.seed(2019-10-09)
ggplot(data = d_edges_filter, 
       aes(
         from_id = from, 
         to_id = to)) +
  geom_net(
    layout.alg = "kamadakawai",
    size = 2, 
    labelon = TRUE, 
    vjust = -0.6, 
    ecolour = "grey60",
    directed =FALSE, 
    fontsize = 3, 
    ealpha = 0.5
    ) +
    theme_net()

10 / 38

Data: Last 4 months of currency USD cross-rates

😨 SO let's try this with cross-currency rates across the globe!

11 / 38

Data: Last 4 months of currency USD cross-rates

😨 SO let's try this with cross-currency rates across the globe!

Data extracted from http://openexchangerates.org/api/historical
R packages jsonlite, processed with tidyverse, lubridate

11 / 38

Data: Last 4 months of currency USD cross-rates

## # A tibble: 6 x 171
##   date         AED   AFN   ALL   AMD   ANG   AOA   ARS   AUD   AWG   AZN   BAM   BBD   BDT
##   <date>     <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 2018-05-14  3.67  71.2  106.  485.  1.79  230.  25.0  1.33  1.79  1.70  1.63     2  84.7
## 2 2018-05-15  3.67  71.2  107.  485.  1.80  230.  24.1  1.34  1.79  1.70  1.64     2  84.8
## 3 2018-05-16  3.67  71.0  108.  484.  1.80  232.  24.3  1.33  1.79  1.70  1.66     2  84.8
## 4 2018-05-17  3.67  71.0  108.  483.  1.80  233.  24.3  1.33  1.79  1.70  1.66     2  84.8
## 5 2018-05-18  3.67  71.0  108.  483.  1.80  233.  24.4  1.33  1.79  1.70  1.66     2  84.8
## 6 2018-05-19  3.67  70.9  108.  482.  1.79  233.  24.4  1.33  1.79  1.70  1.66     2  84.8
## # … with 157 more variables: BGN <dbl>, BHD <dbl>, BIF <dbl>, BMD <dbl>, BND <dbl>,
## #   BOB <dbl>, BRL <dbl>, BSD <dbl>, BTC <dbl>, BTN <dbl>, BWP <dbl>, BYN <dbl>,
## #   BZD <dbl>, CAD <dbl>, CDF <dbl>, CHF <dbl>, CLF <dbl>, CLP <dbl>, CNH <dbl>,
## #   CNY <dbl>, COP <dbl>, CRC <dbl>, CUC <dbl>, CUP <dbl>, CVE <dbl>, CZK <dbl>,
## #   DJF <dbl>, DKK <dbl>, DOP <dbl>, DZD <dbl>, EGP <dbl>, ERN <dbl>, ETB <dbl>,
## #   EUR <dbl>, FJD <dbl>, FKP <dbl>, GBP <dbl>, GEL <dbl>, GGP <dbl>, GHS <dbl>,
## #   GIP <dbl>, GMD <dbl>, GNF <dbl>, GTQ <dbl>, GYD <dbl>, HKD <dbl>, HNL <dbl>,
## #   HRK <dbl>, HTG <dbl>, HUF <dbl>, IDR <dbl>, ILS <dbl>, IMP <dbl>, INR <dbl>,
## #   IQD <dbl>, IRR <dbl>, ISK <dbl>, JEP <dbl>, JMD <dbl>, JOD <dbl>, JPY <dbl>,
## #   KES <dbl>, KGS <dbl>, KHR <dbl>, KMF <dbl>, KPW <dbl>, KRW <dbl>, KWD <dbl>,
## #   KYD <dbl>, KZT <dbl>, LAK <dbl>, LBP <dbl>, LKR <dbl>, LRD <dbl>, LSL <dbl>,
## #   LYD <dbl>, MAD <dbl>, MDL <dbl>, MGA <dbl>, MKD <dbl>, MMK <dbl>, MNT <dbl>,
## #   MOP <dbl>, MRO <dbl>, MRU <dbl>, MUR <dbl>, MVR <dbl>, MWK <dbl>, MXN <dbl>,
## #   MYR <dbl>, MZN <dbl>, NAD <dbl>, NGN <dbl>, NIO <dbl>, NOK <dbl>, NPR <dbl>,
## #   NZD <dbl>, OMR <dbl>, PAB <dbl>, PEN <dbl>, …

12 / 38

Data: Last 4 months of currency USD cross-rates

13 / 38

Your turn: rstudio cloud

Make some plots (or google) to answer these questions

Is the NZD more similar to AUD, EUR, or JPY? (What currency is NZD?)
Is SGD more similar to AUD, EUR, or JPY? (What currency is SGD?)
How many currencies are there in the British Isles?

05:00

14 / 38

15 / 38

Pre-processing data: Keep only currencies that change over the periodSome currencies don't change very much. 
These should be filtered from the analysis, because in a study of currency movement, if it doesn't move then there is nothing more to be said. 

16 / 38

Pre-processing data: Keep only currencies that change over the period

To filter out these currencies we use a statistic called coefficient of variation:

$C o e f V a r i a t i o n = \frac{σ}{μ}$

17 / 38

Pre-processing data: Keep only currencies that change over the period

To filter out these currencies we use a statistic called coefficient of variation:

$C o e f V a r i a t i o n = \frac{σ}{μ}$

Measures standard deviation of currency relative to the mean.
For high means, we expect a currency to change more.
That is, relatively the standard deviation would be larger to consider it to be changing.

17 / 38

cv <- function(x){
  sd(x)/mean(x)
}
rates %>% select(-date) %>%
  summarise_all(funs(cv))

## # A tibble: 1 x 170
##       AED    AFN     ALL     AMD    ANG    AOA   ARS    AUD     AWG     AZN    BAM   BBD
##     <dbl>  <dbl>   <dbl>   <dbl>  <dbl>  <dbl> <dbl>  <dbl>   <dbl>   <dbl>  <dbl> <dbl>
## 1 2.22e-5 0.0111 0.00823 0.00276 0.0120 0.0535 0.116 0.0151 0.00136 3.28e-4 0.0106     0
## # … with 158 more variables: BDT <dbl>, BGN <dbl>, BHD <dbl>, BIF <dbl>, BMD <dbl>,
## #   BND <dbl>, BOB <dbl>, BRL <dbl>, BSD <dbl>, BTC <dbl>, BTN <dbl>, BWP <dbl>,
## #   BYN <dbl>, BZD <dbl>, CAD <dbl>, CDF <dbl>, CHF <dbl>, CLF <dbl>, CLP <dbl>,
## #   CNH <dbl>, CNY <dbl>, COP <dbl>, CRC <dbl>, CUC <dbl>, CUP <dbl>, CVE <dbl>,
## #   CZK <dbl>, DJF <dbl>, DKK <dbl>, DOP <dbl>, DZD <dbl>, EGP <dbl>, ERN <dbl>,
## #   ETB <dbl>, EUR <dbl>, FJD <dbl>, FKP <dbl>, GBP <dbl>, GEL <dbl>, GGP <dbl>,
## #   GHS <dbl>, GIP <dbl>, GMD <dbl>, GNF <dbl>, GTQ <dbl>, GYD <dbl>, HKD <dbl>,
## #   HNL <dbl>, HRK <dbl>, HTG <dbl>, HUF <dbl>, IDR <dbl>, ILS <dbl>, IMP <dbl>,
## #   INR <dbl>, IQD <dbl>, IRR <dbl>, ISK <dbl>, JEP <dbl>, JMD <dbl>, JOD <dbl>,
## #   JPY <dbl>, KES <dbl>, KGS <dbl>, KHR <dbl>, KMF <dbl>, KPW <dbl>, KRW <dbl>,
## #   KWD <dbl>, KYD <dbl>, KZT <dbl>, LAK <dbl>, LBP <dbl>, LKR <dbl>, LRD <dbl>,
## #   LSL <dbl>, LYD <dbl>, MAD <dbl>, MDL <dbl>, MGA <dbl>, MKD <dbl>, MMK <dbl>,
## #   MNT <dbl>, MOP <dbl>, MRO <dbl>, MRU <dbl>, MUR <dbl>, MVR <dbl>, MWK <dbl>,
## #   MXN <dbl>, MYR <dbl>, MZN <dbl>, NAD <dbl>, NGN <dbl>, NIO <dbl>, NOK <dbl>,
## #   NPR <dbl>, NZD <dbl>, OMR <dbl>, PAB <dbl>, …

18 / 38

cv <- function(x){
  sd(x)/mean(x)
}
rates %>% select(-date) %>%
  summarise_all(funs(cv)) %>%
  gather(key = curr, 
         value = cv)

## # A tibble: 170 x 2
##    curr         cv
##    <chr>     <dbl>
##  1 AED   0.0000222
##  2 AFN   0.0111   
##  3 ALL   0.00823  
##  4 AMD   0.00276  
##  5 ANG   0.0120   
##  6 AOA   0.0535   
##  7 ARS   0.116    
##  8 AUD   0.0151   
##  9 AWG   0.00136  
## 10 AZN   0.000328 
## # … with 160 more rows

19 / 38

cv <- function(x){
  sd(x)/mean(x)
}
rates %>% select(-date) %>%
  summarise_all(funs(cv)) %>%
  gather(key = curr, 
         value = cv) %>%
  filter(cv > 0.0027)

## # A tibble: 130 x 2
##    curr       cv
##    <chr>   <dbl>
##  1 AFN   0.0111 
##  2 ALL   0.00823
##  3 AMD   0.00276
##  4 ANG   0.0120 
##  5 AOA   0.0535 
##  6 ARS   0.116  
##  7 AUD   0.0151 
##  8 BAM   0.0106 
##  9 BDT   0.00404
## 10 BGN   0.00932
## # … with 120 more rows

20 / 38

# Compute coefficient of variation. We will only analyse 
# currencies that have changes substantially over this time.
# Dates dropped
cv <- function(x){
  sd(x)/mean(x)
}
rates_sum <- rates %>% select(-date) %>%
  summarise_all(funs(cv)) %>%
  gather(key = curr, 
         value = cv) %>%
  filter(cv > 0.0027)
rates_sub <- select(rates, rates_sum$curr)
head(rates_sub)

## # A tibble: 6 x 130
##     AFN   ALL   AMD   ANG   AOA   ARS   AUD   BAM   BDT   BGN   BIF   BND   BRL      BTC
##   <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>    <dbl>
## 1  71.2  106.  485.  1.79  230.  25.0  1.33  1.63  84.7  1.64 1767.  1.33  3.62 0.000115
## 2  71.2  107.  485.  1.80  230.  24.1  1.34  1.64  84.8  1.65 1773.  1.33  3.65 0.000118
## 3  71.0  108.  484.  1.80  232.  24.3  1.33  1.66  84.8  1.66 1759.  1.33  3.68 0.000120
## 4  71.0  108.  483.  1.80  233.  24.3  1.33  1.66  84.8  1.66 1759.  1.33  3.70 0.000124
## 5  71.0  108.  483.  1.80  233.  24.4  1.33  1.66  84.8  1.66 1762.  1.33  3.74 0.000121
## 6  70.9  108.  482.  1.79  233.  24.4  1.33  1.66  84.8  1.66 1761.  1.33  3.74 0.000121
## # … with 116 more variables: BTN <dbl>, BWP <dbl>, BYN <dbl>, CAD <dbl>, CDF <dbl>,
## #   CHF <dbl>, CLF <dbl>, CLP <dbl>, CNH <dbl>, CNY <dbl>, COP <dbl>, CRC <dbl>,
## #   CVE <dbl>, CZK <dbl>, DKK <dbl>, DOP <dbl>, DZD <dbl>, ERN <dbl>, ETB <dbl>,
## #   EUR <dbl>, FJD <dbl>, FKP <dbl>, GBP <dbl>, GEL <dbl>, GGP <dbl>, GHS <dbl>,
## #   GIP <dbl>, GMD <dbl>, GNF <dbl>, GTQ <dbl>, GYD <dbl>, HNL <dbl>, HRK <dbl>,
## #   HTG <dbl>, HUF <dbl>, IDR <dbl>, ILS <dbl>, IMP <dbl>, INR <dbl>, IRR <dbl>,
## #   ISK <dbl>, JEP <dbl>, JMD <dbl>, JPY <dbl>, KES <dbl>, KMF <dbl>, KRW <dbl>,
## #   KZT <dbl>, LAK <dbl>, LKR <dbl>, LRD <dbl>, LSL <dbl>, LYD <dbl>, MAD <dbl>,
## #   MDL <dbl>, MGA <dbl>, MKD <dbl>, MMK <dbl>, MNT <dbl>, MRU <dbl>, MUR <dbl>,
## #   MVR <dbl>, MWK <dbl>, MXN <dbl>, MYR <dbl>, MZN <dbl>, NAD <dbl>, NIO <dbl>,
## #   NOK <dbl>, NPR <dbl>, NZD <dbl>, PEN <dbl>, PGK <dbl>, PHP <dbl>, PKR <dbl>,
## #   PLN <dbl>, PYG <dbl>, RON <dbl>, RSD <dbl>, RUB <dbl>, RWF <dbl>, SBD <dbl>,
## #   SCR <dbl>, SEK <dbl>, SGD <dbl>, SHP <dbl>, SLL <dbl>, SOS <dbl>, STD <dbl>,
## #   STN <dbl>, SZL <dbl>, THB <dbl>, TJS <dbl>, TMT <dbl>, TND <dbl>, TOP <dbl>,
## #   TRY <dbl>, TWD <dbl>, UAH <dbl>, UGX <dbl>, …

21 / 38

Remove currencies that are not currencies

Some of the currencies ... aren't really currencies. Google these ones: XAG, XDR, XPT - what are they?

02:00

# Remove non-currencies
rates_dropped <- rates_sub %>% select(-ALL, -XAG, -XDR, -XPT)

22 / 38

XAG is Gold XPT is Platinum XDR is special drawing rights

Standardize the currencies

To examine overall trend regardless of actual USD cross rate, standardise the values to have mean 0 and standard deviation 1.

scale01 <- function(x) (x-mean(x))/sd(x)
rates_scaled <-  rates_dropped %>%
  mutate_all(funs(scale01))

23 / 38

What is `mutate_all()`?

Instead of:

rates_dropped %>%
  mutate(AFN = scale01(AFN),
         AMD = scale01(AMD),
         ANG = scale01(ANG),
         ...
         ...)

24 / 38

What is `mutate_all()`?

We can write:

rates_scaled <-  rates_dropped %>%
  mutate_all(funs(scale01))

25 / 38

These are called scoped variants - and there are more:*_if = Do this thing if some condition is met
*_at = Do this thing at these select variables
*_all = Do this thing for all variables

26 / 38

Example: `mutate_if()`

iris %>% mutate_if(is.numeric, scale01)

##     Sepal.Length Sepal.Width Petal.Length   Petal.Width    Species
## 1    -0.89767388  1.01560199  -1.33575163 -1.3110521482     setosa
## 2    -1.13920048 -0.13153881  -1.33575163 -1.3110521482     setosa
## 3    -1.38072709  0.32731751  -1.39239929 -1.3110521482     setosa
## 4    -1.50149039  0.09788935  -1.27910398 -1.3110521482     setosa
## 5    -1.01843718  1.24503015  -1.33575163 -1.3110521482     setosa
## 6    -0.53538397  1.93331463  -1.16580868 -1.0486667950     setosa
## 7    -1.50149039  0.78617383  -1.33575163 -1.1798594716     setosa
## 8    -1.01843718  0.78617383  -1.27910398 -1.3110521482     setosa
## 9    -1.74301699 -0.36096697  -1.33575163 -1.3110521482     setosa
## 10   -1.13920048  0.09788935  -1.27910398 -1.4422448248     setosa
## 11   -0.53538397  1.47445831  -1.27910398 -1.3110521482     setosa
## 12   -1.25996379  0.78617383  -1.22245633 -1.3110521482     setosa
## 13   -1.25996379 -0.13153881  -1.33575163 -1.4422448248     setosa
## 14   -1.86378030 -0.13153881  -1.50569459 -1.4422448248     setosa
## 15   -0.05233076  2.16274279  -1.44904694 -1.3110521482     setosa
## 16   -0.17309407  3.08045544  -1.27910398 -1.0486667950     setosa
## 17   -0.53538397  1.93331463  -1.39239929 -1.0486667950     setosa
## 18   -0.89767388  1.01560199  -1.33575163 -1.1798594716     setosa
## 19   -0.17309407  1.70388647  -1.16580868 -1.1798594716     setosa
## 20   -0.89767388  1.70388647  -1.27910398 -1.1798594716     setosa
## 21   -0.53538397  0.78617383  -1.16580868 -1.3110521482     setosa
## 22   -0.89767388  1.47445831  -1.27910398 -1.0486667950     setosa
## 23   -1.50149039  1.24503015  -1.56234224 -1.3110521482     setosa
## 24   -0.89767388  0.55674567  -1.16580868 -0.9174741184     setosa
## 25   -1.25996379  0.78617383  -1.05251337 -1.3110521482     setosa
## 26   -1.01843718 -0.13153881  -1.22245633 -1.3110521482     setosa
## 27   -1.01843718  0.78617383  -1.22245633 -1.0486667950     setosa
## 28   -0.77691058  1.01560199  -1.27910398 -1.3110521482     setosa
## 29   -0.77691058  0.78617383  -1.33575163 -1.3110521482     setosa
## 30   -1.38072709  0.32731751  -1.22245633 -1.3110521482     setosa
## 31   -1.25996379  0.09788935  -1.22245633 -1.3110521482     setosa
## 32   -0.53538397  0.78617383  -1.27910398 -1.0486667950     setosa
## 33   -0.77691058  2.39217095  -1.27910398 -1.4422448248     setosa
## 34   -0.41462067  2.62159911  -1.33575163 -1.3110521482     setosa
## 35   -1.13920048  0.09788935  -1.27910398 -1.3110521482     setosa
## 36   -1.01843718  0.32731751  -1.44904694 -1.3110521482     setosa
## 37   -0.41462067  1.01560199  -1.39239929 -1.3110521482     setosa
## 38   -1.13920048  1.24503015  -1.33575163 -1.4422448248     setosa
## 39   -1.74301699 -0.13153881  -1.39239929 -1.3110521482     setosa
## 40   -0.89767388  0.78617383  -1.27910398 -1.3110521482     setosa
## 41   -1.01843718  1.01560199  -1.39239929 -1.1798594716     setosa
## 42   -1.62225369 -1.73753594  -1.39239929 -1.1798594716     setosa
## 43   -1.74301699  0.32731751  -1.39239929 -1.3110521482     setosa
## 44   -1.01843718  1.01560199  -1.22245633 -0.7862814418     setosa
## 45   -0.89767388  1.70388647  -1.05251337 -1.0486667950     setosa
## 46   -1.25996379 -0.13153881  -1.33575163 -1.1798594716     setosa
## 47   -0.89767388  1.70388647  -1.22245633 -1.3110521482     setosa
## 48   -1.50149039  0.32731751  -1.33575163 -1.3110521482     setosa
## 49   -0.65614727  1.47445831  -1.27910398 -1.3110521482     setosa
## 50   -1.01843718  0.55674567  -1.33575163 -1.3110521482     setosa
## 51    1.39682886  0.32731751   0.53362088  0.2632599711 versicolor
## 52    0.67224905  0.32731751   0.42032558  0.3944526477 versicolor
## 53    1.27606556  0.09788935   0.64691619  0.3944526477 versicolor
## 54   -0.41462067 -1.73753594   0.13708732  0.1320672944 versicolor
## 55    0.79301235 -0.59039513   0.47697323  0.3944526477 versicolor
## 56   -0.17309407 -0.59039513   0.42032558  0.1320672944 versicolor
## 57    0.55148575  0.55674567   0.53362088  0.5256453243 versicolor
## 58   -1.13920048 -1.50810778  -0.25944625 -0.2615107354 versicolor
## 59    0.91377565 -0.36096697   0.47697323  0.1320672944 versicolor
## 60   -0.77691058 -0.81982329   0.08043967  0.2632599711 versicolor
## 61   -1.01843718 -2.42582042  -0.14615094 -0.2615107354 versicolor
## 62    0.06843254 -0.13153881   0.25038262  0.3944526477 versicolor
## 63    0.18919584 -1.96696410   0.13708732 -0.2615107354 versicolor
## 64    0.30995914 -0.36096697   0.53362088  0.2632599711 versicolor
## 65   -0.29385737 -0.36096697  -0.08950329  0.1320672944 versicolor
## 66    1.03453895  0.09788935   0.36367793  0.2632599711 versicolor
## 67   -0.29385737 -0.13153881   0.42032558  0.3944526477 versicolor
## 68   -0.05233076 -0.81982329   0.19373497 -0.2615107354 versicolor
## 69    0.43072244 -1.96696410   0.42032558  0.3944526477 versicolor
## 70   -0.29385737 -1.27867961   0.08043967 -0.1303180588 versicolor
## 71    0.06843254  0.32731751   0.59026853  0.7880306775 versicolor
## 72    0.30995914 -0.59039513   0.13708732  0.1320672944 versicolor
## 73    0.55148575 -1.27867961   0.64691619  0.3944526477 versicolor
## 74    0.30995914 -0.59039513   0.53362088  0.0008746178 versicolor
## 75    0.67224905 -0.36096697   0.30703027  0.1320672944 versicolor
## 76    0.91377565 -0.13153881   0.36367793  0.2632599711 versicolor
## 77    1.15530226 -0.59039513   0.59026853  0.2632599711 versicolor
## 78    1.03453895 -0.13153881   0.70356384  0.6568380009 versicolor
## 79    0.18919584 -0.36096697   0.42032558  0.3944526477 versicolor
## 80   -0.17309407 -1.04925145  -0.14615094 -0.2615107354 versicolor
## 81   -0.41462067 -1.50810778   0.02379201 -0.1303180588 versicolor
## 82   -0.41462067 -1.50810778  -0.03285564 -0.2615107354 versicolor
## 83   -0.05233076 -0.81982329   0.08043967  0.0008746178 versicolor
## 84    0.18919584 -0.81982329   0.76021149  0.5256453243 versicolor
## 85   -0.53538397 -0.13153881   0.42032558  0.3944526477 versicolor
## 86    0.18919584  0.78617383   0.42032558  0.5256453243 versicolor
## 87    1.03453895  0.09788935   0.53362088  0.3944526477 versicolor
## 88    0.55148575 -1.73753594   0.36367793  0.1320672944 versicolor
## 89   -0.29385737 -0.13153881   0.19373497  0.1320672944 versicolor
## 90   -0.41462067 -1.27867961   0.13708732  0.1320672944 versicolor
## 91   -0.41462067 -1.04925145   0.36367793  0.0008746178 versicolor
## 92    0.30995914 -0.13153881   0.47697323  0.2632599711 versicolor
## 93   -0.05233076 -1.04925145   0.13708732  0.0008746178 versicolor
## 94   -1.01843718 -1.73753594  -0.25944625 -0.2615107354 versicolor
## 95   -0.29385737 -0.81982329   0.25038262  0.1320672944 versicolor
## 96   -0.17309407 -0.13153881   0.25038262  0.0008746178 versicolor
## 97   -0.17309407 -0.36096697   0.25038262  0.1320672944 versicolor
## 98    0.43072244 -0.36096697   0.30703027  0.1320672944 versicolor
## 99   -0.89767388 -1.27867961  -0.42938920 -0.1303180588 versicolor
## 100  -0.17309407 -0.59039513   0.19373497  0.1320672944 versicolor
## 101   0.55148575  0.55674567   1.27004036  1.7063794137  virginica
## 102  -0.05233076 -0.81982329   0.76021149  0.9192233541  virginica
## 103   1.51759216 -0.13153881   1.21339271  1.1816087073  virginica
## 104   0.55148575 -0.36096697   1.04344975  0.7880306775  virginica
## 105   0.79301235 -0.13153881   1.15674505  1.3128013839  virginica
## 106   2.12140867 -0.13153881   1.60992627  1.1816087073  virginica
## 107  -1.13920048 -1.27867961   0.42032558  0.6568380009  virginica
## 108   1.75911877 -0.36096697   1.43998331  0.7880306775  virginica
## 109   1.03453895 -1.27867961   1.15674505  0.7880306775  virginica
## 110   1.63835547  1.24503015   1.32668801  1.7063794137  virginica
## 111   0.79301235  0.32731751   0.76021149  1.0504160307  virginica
## 112   0.67224905 -0.81982329   0.87350679  0.9192233541  virginica
## 113   1.15530226 -0.13153881   0.98680210  1.1816087073  virginica
## 114  -0.17309407 -1.27867961   0.70356384  1.0504160307  virginica
## 115  -0.05233076 -0.59039513   0.76021149  1.5751867371  virginica
## 116   0.67224905  0.32731751   0.87350679  1.4439940605  virginica
## 117   0.79301235 -0.13153881   0.98680210  0.7880306775  virginica
## 118   2.24217198  1.70388647   1.66657392  1.3128013839  virginica
## 119   2.24217198 -1.04925145   1.77986923  1.4439940605  virginica
## 120   0.18919584 -1.96696410   0.70356384  0.3944526477  virginica
## 121   1.27606556  0.32731751   1.10009740  1.4439940605  virginica
## 122  -0.29385737 -0.59039513   0.64691619  1.0504160307  virginica
## 123   2.24217198 -0.59039513   1.66657392  1.0504160307  virginica
## 124   0.55148575 -0.81982329   0.64691619  0.7880306775  virginica
## 125   1.03453895  0.55674567   1.10009740  1.1816087073  virginica
## 126   1.63835547  0.32731751   1.27004036  0.7880306775  virginica
## 127   0.43072244 -0.59039513   0.59026853  0.7880306775  virginica
## 128   0.30995914 -0.13153881   0.64691619  0.7880306775  virginica
## 129   0.67224905 -0.59039513   1.04344975  1.1816087073  virginica
## 130   1.63835547 -0.13153881   1.15674505  0.5256453243  virginica
## 131   1.87988207 -0.59039513   1.32668801  0.9192233541  virginica
## 132   2.48369858  1.70388647   1.49663097  1.0504160307  virginica
## 133   0.67224905 -0.59039513   1.04344975  1.3128013839  virginica
## 134   0.55148575 -0.59039513   0.76021149  0.3944526477  virginica
## 135   0.30995914 -1.04925145   1.04344975  0.2632599711  virginica
## 136   2.24217198 -0.13153881   1.32668801  1.4439940605  virginica
## 137   0.55148575  0.78617383   1.04344975  1.5751867371  virginica
## 138   0.67224905  0.09788935   0.98680210  0.7880306775  virginica
## 139   0.18919584 -0.13153881   0.59026853  0.7880306775  virginica
## 140   1.27606556  0.09788935   0.93015445  1.1816087073  virginica
## 141   1.03453895  0.09788935   1.04344975  1.5751867371  virginica
## 142   1.27606556  0.09788935   0.76021149  1.4439940605  virginica
## 143  -0.05233076 -0.81982329   0.76021149  0.9192233541  virginica
## 144   1.15530226  0.32731751   1.21339271  1.4439940605  virginica
## 145   1.03453895  0.55674567   1.10009740  1.7063794137  virginica
## 146   1.03453895 -0.13153881   0.81685914  1.4439940605  virginica
## 147   0.55148575 -1.27867961   0.70356384  0.9192233541  virginica
## 148   0.79301235 -0.13153881   0.81685914  1.0504160307  virginica
## 149   0.43072244  0.78617383   0.93015445  1.4439940605  virginica
## 150   0.06843254 -0.13153881   0.76021149  0.7880306775  virginica

27 / 38

Example: `mutate_at()`

iris %>% mutate_at(vars(Sepal.Width, Sepal.Length), scale01)

##     Sepal.Length Sepal.Width Petal.Length Petal.Width    Species
## 1    -0.89767388  1.01560199          1.4         0.2     setosa
## 2    -1.13920048 -0.13153881          1.4         0.2     setosa
## 3    -1.38072709  0.32731751          1.3         0.2     setosa
## 4    -1.50149039  0.09788935          1.5         0.2     setosa
## 5    -1.01843718  1.24503015          1.4         0.2     setosa
## 6    -0.53538397  1.93331463          1.7         0.4     setosa
## 7    -1.50149039  0.78617383          1.4         0.3     setosa
## 8    -1.01843718  0.78617383          1.5         0.2     setosa
## 9    -1.74301699 -0.36096697          1.4         0.2     setosa
## 10   -1.13920048  0.09788935          1.5         0.1     setosa
## 11   -0.53538397  1.47445831          1.5         0.2     setosa
## 12   -1.25996379  0.78617383          1.6         0.2     setosa
## 13   -1.25996379 -0.13153881          1.4         0.1     setosa
## 14   -1.86378030 -0.13153881          1.1         0.1     setosa
## 15   -0.05233076  2.16274279          1.2         0.2     setosa
## 16   -0.17309407  3.08045544          1.5         0.4     setosa
## 17   -0.53538397  1.93331463          1.3         0.4     setosa
## 18   -0.89767388  1.01560199          1.4         0.3     setosa
## 19   -0.17309407  1.70388647          1.7         0.3     setosa
## 20   -0.89767388  1.70388647          1.5         0.3     setosa
## 21   -0.53538397  0.78617383          1.7         0.2     setosa
## 22   -0.89767388  1.47445831          1.5         0.4     setosa
## 23   -1.50149039  1.24503015          1.0         0.2     setosa
## 24   -0.89767388  0.55674567          1.7         0.5     setosa
## 25   -1.25996379  0.78617383          1.9         0.2     setosa
## 26   -1.01843718 -0.13153881          1.6         0.2     setosa
## 27   -1.01843718  0.78617383          1.6         0.4     setosa
## 28   -0.77691058  1.01560199          1.5         0.2     setosa
## 29   -0.77691058  0.78617383          1.4         0.2     setosa
## 30   -1.38072709  0.32731751          1.6         0.2     setosa
## 31   -1.25996379  0.09788935          1.6         0.2     setosa
## 32   -0.53538397  0.78617383          1.5         0.4     setosa
## 33   -0.77691058  2.39217095          1.5         0.1     setosa
## 34   -0.41462067  2.62159911          1.4         0.2     setosa
## 35   -1.13920048  0.09788935          1.5         0.2     setosa
## 36   -1.01843718  0.32731751          1.2         0.2     setosa
## 37   -0.41462067  1.01560199          1.3         0.2     setosa
## 38   -1.13920048  1.24503015          1.4         0.1     setosa
## 39   -1.74301699 -0.13153881          1.3         0.2     setosa
## 40   -0.89767388  0.78617383          1.5         0.2     setosa
## 41   -1.01843718  1.01560199          1.3         0.3     setosa
## 42   -1.62225369 -1.73753594          1.3         0.3     setosa
## 43   -1.74301699  0.32731751          1.3         0.2     setosa
## 44   -1.01843718  1.01560199          1.6         0.6     setosa
## 45   -0.89767388  1.70388647          1.9         0.4     setosa
## 46   -1.25996379 -0.13153881          1.4         0.3     setosa
## 47   -0.89767388  1.70388647          1.6         0.2     setosa
## 48   -1.50149039  0.32731751          1.4         0.2     setosa
## 49   -0.65614727  1.47445831          1.5         0.2     setosa
## 50   -1.01843718  0.55674567          1.4         0.2     setosa
## 51    1.39682886  0.32731751          4.7         1.4 versicolor
## 52    0.67224905  0.32731751          4.5         1.5 versicolor
## 53    1.27606556  0.09788935          4.9         1.5 versicolor
## 54   -0.41462067 -1.73753594          4.0         1.3 versicolor
## 55    0.79301235 -0.59039513          4.6         1.5 versicolor
## 56   -0.17309407 -0.59039513          4.5         1.3 versicolor
## 57    0.55148575  0.55674567          4.7         1.6 versicolor
## 58   -1.13920048 -1.50810778          3.3         1.0 versicolor
## 59    0.91377565 -0.36096697          4.6         1.3 versicolor
## 60   -0.77691058 -0.81982329          3.9         1.4 versicolor
## 61   -1.01843718 -2.42582042          3.5         1.0 versicolor
## 62    0.06843254 -0.13153881          4.2         1.5 versicolor
## 63    0.18919584 -1.96696410          4.0         1.0 versicolor
## 64    0.30995914 -0.36096697          4.7         1.4 versicolor
## 65   -0.29385737 -0.36096697          3.6         1.3 versicolor
## 66    1.03453895  0.09788935          4.4         1.4 versicolor
## 67   -0.29385737 -0.13153881          4.5         1.5 versicolor
## 68   -0.05233076 -0.81982329          4.1         1.0 versicolor
## 69    0.43072244 -1.96696410          4.5         1.5 versicolor
## 70   -0.29385737 -1.27867961          3.9         1.1 versicolor
## 71    0.06843254  0.32731751          4.8         1.8 versicolor
## 72    0.30995914 -0.59039513          4.0         1.3 versicolor
## 73    0.55148575 -1.27867961          4.9         1.5 versicolor
## 74    0.30995914 -0.59039513          4.7         1.2 versicolor
## 75    0.67224905 -0.36096697          4.3         1.3 versicolor
## 76    0.91377565 -0.13153881          4.4         1.4 versicolor
## 77    1.15530226 -0.59039513          4.8         1.4 versicolor
## 78    1.03453895 -0.13153881          5.0         1.7 versicolor
## 79    0.18919584 -0.36096697          4.5         1.5 versicolor
## 80   -0.17309407 -1.04925145          3.5         1.0 versicolor
## 81   -0.41462067 -1.50810778          3.8         1.1 versicolor
## 82   -0.41462067 -1.50810778          3.7         1.0 versicolor
## 83   -0.05233076 -0.81982329          3.9         1.2 versicolor
## 84    0.18919584 -0.81982329          5.1         1.6 versicolor
## 85   -0.53538397 -0.13153881          4.5         1.5 versicolor
## 86    0.18919584  0.78617383          4.5         1.6 versicolor
## 87    1.03453895  0.09788935          4.7         1.5 versicolor
## 88    0.55148575 -1.73753594          4.4         1.3 versicolor
## 89   -0.29385737 -0.13153881          4.1         1.3 versicolor
## 90   -0.41462067 -1.27867961          4.0         1.3 versicolor
## 91   -0.41462067 -1.04925145          4.4         1.2 versicolor
## 92    0.30995914 -0.13153881          4.6         1.4 versicolor
## 93   -0.05233076 -1.04925145          4.0         1.2 versicolor
## 94   -1.01843718 -1.73753594          3.3         1.0 versicolor
## 95   -0.29385737 -0.81982329          4.2         1.3 versicolor
## 96   -0.17309407 -0.13153881          4.2         1.2 versicolor
## 97   -0.17309407 -0.36096697          4.2         1.3 versicolor
## 98    0.43072244 -0.36096697          4.3         1.3 versicolor
## 99   -0.89767388 -1.27867961          3.0         1.1 versicolor
## 100  -0.17309407 -0.59039513          4.1         1.3 versicolor
## 101   0.55148575  0.55674567          6.0         2.5  virginica
## 102  -0.05233076 -0.81982329          5.1         1.9  virginica
## 103   1.51759216 -0.13153881          5.9         2.1  virginica
## 104   0.55148575 -0.36096697          5.6         1.8  virginica
## 105   0.79301235 -0.13153881          5.8         2.2  virginica
## 106   2.12140867 -0.13153881          6.6         2.1  virginica
## 107  -1.13920048 -1.27867961          4.5         1.7  virginica
## 108   1.75911877 -0.36096697          6.3         1.8  virginica
## 109   1.03453895 -1.27867961          5.8         1.8  virginica
## 110   1.63835547  1.24503015          6.1         2.5  virginica
## 111   0.79301235  0.32731751          5.1         2.0  virginica
## 112   0.67224905 -0.81982329          5.3         1.9  virginica
## 113   1.15530226 -0.13153881          5.5         2.1  virginica
## 114  -0.17309407 -1.27867961          5.0         2.0  virginica
## 115  -0.05233076 -0.59039513          5.1         2.4  virginica
## 116   0.67224905  0.32731751          5.3         2.3  virginica
## 117   0.79301235 -0.13153881          5.5         1.8  virginica
## 118   2.24217198  1.70388647          6.7         2.2  virginica
## 119   2.24217198 -1.04925145          6.9         2.3  virginica
## 120   0.18919584 -1.96696410          5.0         1.5  virginica
## 121   1.27606556  0.32731751          5.7         2.3  virginica
## 122  -0.29385737 -0.59039513          4.9         2.0  virginica
## 123   2.24217198 -0.59039513          6.7         2.0  virginica
## 124   0.55148575 -0.81982329          4.9         1.8  virginica
## 125   1.03453895  0.55674567          5.7         2.1  virginica
## 126   1.63835547  0.32731751          6.0         1.8  virginica
## 127   0.43072244 -0.59039513          4.8         1.8  virginica
## 128   0.30995914 -0.13153881          4.9         1.8  virginica
## 129   0.67224905 -0.59039513          5.6         2.1  virginica
## 130   1.63835547 -0.13153881          5.8         1.6  virginica
## 131   1.87988207 -0.59039513          6.1         1.9  virginica
## 132   2.48369858  1.70388647          6.4         2.0  virginica
## 133   0.67224905 -0.59039513          5.6         2.2  virginica
## 134   0.55148575 -0.59039513          5.1         1.5  virginica
## 135   0.30995914 -1.04925145          5.6         1.4  virginica
## 136   2.24217198 -0.13153881          6.1         2.3  virginica
## 137   0.55148575  0.78617383          5.6         2.4  virginica
## 138   0.67224905  0.09788935          5.5         1.8  virginica
## 139   0.18919584 -0.13153881          4.8         1.8  virginica
## 140   1.27606556  0.09788935          5.4         2.1  virginica
## 141   1.03453895  0.09788935          5.6         2.4  virginica
## 142   1.27606556  0.09788935          5.1         2.3  virginica
## 143  -0.05233076 -0.81982329          5.1         1.9  virginica
## 144   1.15530226  0.32731751          5.9         2.3  virginica
## 145   1.03453895  0.55674567          5.7         2.5  virginica
## 146   1.03453895 -0.13153881          5.2         2.3  virginica
## 147   0.55148575 -1.27867961          5.0         1.9  virginica
## 148   0.79301235 -0.13153881          5.2         2.0  virginica
## 149   0.43072244  0.78617383          5.4         2.3  virginica
## 150   0.06843254 -0.13153881          5.1         1.8  virginica

28 / 38

Scoped Variants (if, at, all*) are available for the five verbs:mutate()
filter()
select()
summarise()
group_by()

29 / 38

Standardize the currencies

30 / 38

Compute distances between all pairs of currencies

Euclidean distance is used to compute similarity between all pairs of currencies.

$d_{i j} = \sqrt{\sum_{i = 1}^{t} (C_{1 i} - C_{2 i})^{2}}$

31 / 38

Compute distances between all pairs of currencies

# Compute distance between currencies
# Need to transpose! Turn matrix around, rows <--> columns
rates_dropped_t <- t(rates_dropped) %>% data.frame()
dates_dist <- as.matrix(dist(rates_dropped_t, 
                             diag = TRUE, 
                             upper = TRUE))
colnames(dates_dist) <- as.factor(colnames(rates_dropped))
rownames(dates_dist) <- as.factor(colnames(rates_dropped))
quantile(dates_dist, 
         probs = c(0, 0.25, 0.5, 0.75, 1))

##           0%          25%          50%          75%         100% 
##       0.0000     172.2863    1074.7584   11948.3344 1449147.7492

32 / 38

A note on distance matrices:A distance matrix is the inverse of an association matrix. 
A distance matrix close to 0 means the pair are most similar. 
For an association matrix far from zero means the pair are close. 
Either can be used to generate a network. 

33 / 38

Make the network: Gather the data into long form, and filter based on similarity

Here only the pairs of currencies who are closer than "4" to each other are kept.

d_zero <- d
d_zero_tbl <- d_zero %>% 
  as_tibble() %>%
  mutate(curr1=rownames(d_zero)) %>%
  gather(curr2, dst, -curr1) %>%
  filter(dst<3) %>%
  filter(curr1 != curr2)

34 / 38

Make the network: Gather the data into long form, and filter based on similarity

Here only the pairs of currencies who are closer than "4" to each other are kept.

d_zero_tbl

## # A tibble: 12 x 3
##    curr1 curr2   dst
##    <chr> <chr> <dbl>
##  1 Zili  Meg       1
##  2 Jess  Meg       1
##  3 Zili  Tay       2
##  4 Jess  Tay       1
##  5 Zili  Yat       0
##  6 Jess  Yat       0
##  7 Meg   Zili      1
##  8 Tay   Zili      2
##  9 Yat   Zili      0
## 10 Meg   Jess      1
## 11 Tay   Jess      1
## 12 Yat   Jess      0

35 / 38

Network laid out

# Make network
library(geomnet)
set.seed(10052016)
ggplot(data = d_zero_tbl, 
       aes(
         from_id = curr1, 
         to_id = curr2
         )) +
  geom_net(
    layout.alg = "kamadakawai",
    size = 2, 
    labelon = TRUE, 
    vjust = -0.6, 
    ecolour = "grey60",
    directed = FALSE, 
    fontsize = 3, 
    ealpha = 0.5
    ) +
    theme_net() +
    theme(
      legend.position = "bottom"
      )

36 / 38

Your turnMake a plot of the AUD vs the SGD (using the standardised units). Do they look like they are trending together as suggested by the network?
Finish the rstudio cloud exercise
Reamining time: Ask questions about project / assignment

37 / 38

This work is licensed under a Creative Commons Attribution 4.0 International License.

38 / 38

Announcements

Assignment 3 is due this Friday at 5pm
Project deadlines:
- Deadline 3 (11th October) : Electronic copy of your data, and a page of data description, and cleaning done, or needing to be done.
- Deadline 4 (18th October) : Final version of story board uploaded.
Guest Lecture: 16th October: Dr. James McKeone
Practical exam: 18th October in class at 8am
Final Exam: I will provide a review of exam content

↑, ←, Pg Up, k	Go to previous slide
↓, →, Pg Dn, Space, j	Go to next slide
Home	Go to first slide
End	Go to last slide
Number + Return	Go to specific slide
b / m / f	Toggle blackout / mirrored / fullscreen mode
c	Clone slideshow
p	Toggle presenter mode
t	Restart the presentation timer
?, h	Toggle this help

ETC1010: Data Modelling and Computing

Lecture 9B: More on networks

Dr. Nicholas Tierney & Professor Di Cook

EBS, Monash U.

2019-10-09

Announcements

recap: on network data

Quantitative association matrices

Quantitative association matrices

Quantitative association matrices

Quantitative association matrices

Quantitative association matrices

Quantitative association matrices

Quantitative association matrices: Make the network diagram.

Data: Last 4 months of currency USD cross-rates

Data: Last 4 months of currency USD cross-rates

Data: Last 4 months of currency USD cross-rates

Data: Last 4 months of currency USD cross-rates

Your turn: rstudio cloud

Pre-processing data: Keep only currencies that change over the period

Pre-processing data: Keep only currencies that change over the period

Pre-processing data: Keep only currencies that change over the period

Remove currencies that are not currencies

Standardize the currencies

What is `mutate_all()`?

What is `mutate_all()`?

These are called `scoped variants` - and there are more:

Example: `mutate_if()`

Example: `mutate_at()`

Scoped Variants (if, at, all*) are available for the five verbs:

Standardize the currencies

Compute distances between all pairs of currencies

Compute distances between all pairs of currencies

A note on distance matrices:

Make the network: Gather the data into long form, and filter based on similarity

Make the network: Gather the data into long form, and filter based on similarity

Network laid out

Your turn

Announcements

Help

ETC1010: Data Modelling and Computing

Lecture 9B: More on networks

Dr. Nicholas Tierney & Professor Di Cook

EBS, Monash U.

2019-10-09

Announcements

recap: on network data

Quantitative association matrices

Quantitative association matrices

Quantitative association matrices

Quantitative association matrices

Quantitative association matrices

Quantitative association matrices

Quantitative association matrices: Make the network diagram.

Data: Last 4 months of currency USD cross-rates

Data: Last 4 months of currency USD cross-rates

Data: Last 4 months of currency USD cross-rates

Data: Last 4 months of currency USD cross-rates

Your turn: rstudio cloud

Pre-processing data: Keep only currencies that change over the period

Pre-processing data: Keep only currencies that change over the period

Pre-processing data: Keep only currencies that change over the period

Remove currencies that are not currencies

Standardize the currencies

What is mutate_all()?

What is mutate_all()?

These are called scoped variants - and there are more:

Example: mutate_if()

Example: mutate_at()

Scoped Variants (if, at, all*) are available for the five verbs:

Standardize the currencies

Compute distances between all pairs of currencies

Compute distances between all pairs of currencies

A note on distance matrices:

Make the network: Gather the data into long form, and filter based on similarity

Make the network: Gather the data into long form, and filter based on similarity

Network laid out

Your turn

Share and share alike

Announcements

Help

What is `mutate_all()`?

What is `mutate_all()`?

These are called `scoped variants` - and there are more:

Example: `mutate_if()`

Example: `mutate_at()`