NYCJUG/2010-02-09/LargeDatasetsStatsCase0

From J Wiki
Jump to navigation Jump to search

Some statistics, shown as histograms, on random datasets generated to test parallel programs.

Cashflows

Let’s take a look at the distribution of cashflows to see if they look plausible. The histogram function plotHistoMult can be found here.

   load 'mystats'                            NB. My histogram function
   ss=. 'Cashflows (K)' plotHistoMulti ,cfs%1000 [ PCT=: 0

CashflowDistrib.png These look fairly evenly distributed across the range. A more realistic series might be concentrated more on the lower end but this should be good enough for our purposes as one of our goals is simplicity of generation. So, let's also look at the differences within each series to ensure they don’t fluctuate wildly.

   ss=. 'Cashflow Differences/Series (K)' plotHistoMulti ,(>./-<./)"1 cfs%1000

CashflowDiffs.png Not too bad but what does this look like as a percentage of the mid-point?

   pctdif=. ((>./-<./)%[:-:(>./+<./))"1 cfs
   ss=. 'Cashflow %Difference/Series' plotHistoMulti pctdif [ PCT=: 1

CashflowPctDiffs.png

Interest Rates

   ss=. 'Interest Rates' plotHistoMulti ,irp [ PCT=: 1

InterestRatesDistrib.png

Present Values

viewmat pvs

ViewPVs.jpg