JhonatanHern

const fs = require('fs').promises

const moduleFolders = ['defs', 'lib', 'rpc', 'storage', 'storage_consts']

const findOffenders = async fileArray => {
  const offenders = []
  for (let index = 0; index < fileArray.length; index++) {
    const file = fileArray[index]
    const data = (await fs.readFile(file)).toString()
    if (data[0]!=='/'){

JhonatanHern

const {
  getDNA,
  buildConfig,
  buildRunner,
} = require('../init')

const runner = buildRunner()

const config = buildConfig({
  proposal: getDNA('proposal'),

JhonatanHern

##
#
# System mutations entrypoint
#
# Conventions for all operations are as follows:
# - input record data is provided in named sub-fields of the request & response
#   - the sub-field ID is the lower camelCase name of the record type
#
# CREATE:
# - fields which are required by the record are required in input parameters

JhonatanHern

#https://archive.ics.uci.edu/ml/datasets/Post-Operative+Patient
library(neuralnet)
library(dplyr)

read.table('post-operative.data',sep=',') -> po
randomized_data <- sample_frac(po,1)

training <- randomized_data[1:79,]
testd <- randomized_data[80:90,]

JhonatanHern

- - - - - - - - - - - - - - - - - Particionamiento
Se suelen usar distancias estilo euclideana, Manhattan o Minkowski.
Se generan particiones (asignar etiquetas a un area) similares a "nubes"
Cada nube tiene un "centroide"

Se usan dos distancias:
Distancia Inte-cluster
Distancia Intra-cluster

Un algoritmo usado es el de k-means. Este requiere una función para sacar el promedio de un conjunto de observaciones y 

JhonatanHern

Clusterización:

Detección de grupos en datasets mediante distancias multidimensionales

Tipos de algoritmos para cálculo de distancia según tipo de variables:

------------------------------ Numérica (con números)
( (x11-x12)**2 + (x21-x22)**2 + ... + (xn1-xn2)**2 )**(1/2) # Euclideana

|x11-x12| + ... + |xn1-xn2|                                 # Manhattan

JhonatanHern

#[macro_use]
extern crate hdk;
extern crate serde;
#[macro_use]
extern crate serde_derive;
extern crate serde_json;
#[macro_use]


extern crate holochain_json_derive;

JhonatanHern

const path = require('path')
const tape = require('tape')

const { Diorama, tapeExecutor, backwardCompatibilityMiddleware } = require('@holochain/diorama')

process.on('unhandledRejection', error => {
  // Will print "unhandledRejection err is not defined"
  console.error('got unhandledRejection:', error);
})

JhonatanHern

library(neuralnet)
library(dplyr)

random.iris <- sample_frac(iris,1)

miris <- random.iris[1:105,]
testd <- random.iris[106:150,]

miris <- mutate(miris,
       setosa = Species == "setosa",

JhonatanHern

library(neuralnet)

# creating training data set
TKS=   c(20,10,30,20,80,30,90,99,99,40,12,32)
CSS=   c(90,20,40,50,50,80,90,99,99,10,20,30)
Placed=c(1 ,0 ,0 ,0 ,1 ,1 ,1 ,1 ,1 ,0 ,0 ,0)

df=data.frame(TKS,CSS,Placed)

nn=neuralnet(Placed~TKS+CSS,data=df, hidden=3, act.fct = "logistic", linear.output = FALSE)