How to work with time series

function window extracts time subperiod from start till finish!

plot2(window(season2016,start=c(2017,1)),season2017,"2016","2017","weeks","UAH")

How to bring dnorm to %

height distribution

s=seq(1.4,2.4,0.01) #height is between 1.4 and 2.4 m
d=dnorm(s,2,0.1) #height is distributed normally with mean = 2 m and sd =0.1
plot(s,d,type="l")

To switch to %:

ddnorm=function(x,m,s){d=dnorm(x,m,s);dd=d/sum(d);return(dd)}

plot(seq(1.4,2.4,0.01),ddnorm(seq(1.4,2.4,0.01),2,0.1),type="l")

sum(ddnorm(seq(1.4,2.4,0.01),2,0.1)) #test on 100%

[1] 1

Check two vectors on equality of unique elements

x=c(5,5,5,5,5,5)
y=c(5,5)
z=c(3,5,5)

unique(x)
[1] 5
unique(y)
[1] 5
unique(z)
[1] 3 5

unique(x)==unique(y)
[1] TRUE
unique(x)==unique(z) #if sets are of different length we need to use all function
[1] FALSE  TRUE
all(unique(x)==unique(z))
[1] FALSE
all(unique(x)==unique(y))
[1] TRUE

Check vectors on equality

a=c(3,4)
b=c(3,5)
d=c(3,4)

a==b
[1]  TRUE FALSE

check 1
all(a==b)
[1] FALSE
all(a==d)
[1] TRUE

check 2
all.equal(a,b)
[1] "Mean relative difference: 0.25"
#if we need TRUE or FALSE
isTRUE(all.equal(a,b))
[1] FALSE
isTRUE(all.equal(a,d))
[1] TRUE

Normal Distribution Plot

m=mean(testData)
s=sd(testData)
plot(dnorm(seq(0,8,1),m,s),type="l")

Test of Normal Distribution Shapiro-Wilk

shapiro.test(rnorm(500))

Shapiro-Wilk normality test

data:  rnorm(500)
W = 0.99677, p-value = 0.4218

Since the p-value is > 0.05, it is accepted the dataset is normally distributed

Advices

https://medium.com/@rchang/advice-for-new-and-junior-data-scientists-2ab02396cf5b
https://towardsdatascience.com/top-6-errors-novice-machine-learning-engineers-make-e82273d394db
https://www.kaggle.com/surveys/2017