CODEWITHSUNDEEP
rregressionmidasr
Jeweller89
Jeweller89

Reputation: 19

Mixed data sampling in R (midas_r)

I have the following code where I have a weekly time series (variable x) which I would like to use in order to forecast my monthly time series (variable y).

So basically I want to forecast the current month's data (variable y) with either 1,2,3 or all 4 weeks (variable New_value) in the current month.

However, I am not sure if I am using the correct lags (I think I am) but moreover I am not sure how to interpret the starting values in the midas_r function (start = list() ).

Any help would be much appreciated.

######################
# MIDAS REGRESSION
####################
x <- structure(c(1.19, 1.24 , 1.67 , 1.67 , 1.55 , 1.67 , 1.39 , 2.01 , 2.14 , 1.71 , 1.59 , 1.49 , 1.68 , -0.37 , -0.44 , -7.87 , -7.79 , -31.22 , -31.05 , -30.47 , -35.53 , -25.48 , -25.9 , -19.03 , -16.33 , 
                 10.09 , 13.19 , 13.31 , 16.85 , 14.58 , 14.78 , 14.62 , 15.27 , 15.58 , 15.63 , 14.27 , 14.09 , 4.82 , 3.55 , 3.46 , 3.24 , 2.86 , 2.86 , 2.86 , 2.82),
               .Tsp = c(2020, 2020.846154, 52), class = "ts")
x <- diff(x)

y <- structure(c(2.30, 2.64 , 2.77 , 2.83 , -43.91 , 12.32 , 26.68 , 12.06 , 10.08 , 12.01 , 4.71 , 3.85),
               .Tsp = c(2020, 2020.91667, 12), class = "ts")
y <- diff(y)

trend <- c(1:length(y))

#RUNNING THE MIDAS REGRESSION
reg <- midas_r(y ~ mls(y, 1, 1) + mls(x, 4:7, m = 4, nealmon), start = list(x = c(1, 1, -1, -1)))
summary(reg)
hAh_test(reg)

#forecast(reg, newdata = list(y = c(NA), x =c(rep(NA, 4))))

#FORECAST WITH WEEKLY VALUES
reg <- midas_r(y ~ mls(y, 1, 1) + mls(x, 3:7, m = 4, nealmon), start = list(x = c(1, 1, -1, -1)))
new_value <- 2.52
#new_value <- c(2.52, 3.12)
forecast(reg, newdata = list(x = c(new_value, rep(NA, 4-length(new_value)))))

Upvotes: -1

Views: 358

Answers (1)

Jack Fleeting
Jack Fleeting

Reputation: 24930

Looks like you have a problem with indentation, plus some unnecessary complexity.

Try something like this and see if it work:

for i in range(1, 20):
    url = 'https://www.boliga.dk/salg/resultater?propertyType=4&street=&municipality=326&salesDateMin=2020&page=' + str(i)
    page = requests.get(url)
    soup = BeautifulSoup(page.text, 'html.parser')    
    table = soup.find('table', {'class': 'table generic-table m-0 mb-3'})
    df = pd.read_html(str(table))[0]
    df = pd.DataFrame(df) #if it all works otherwise, try running the code w/out this line; it may be unnecessary
    df

Upvotes: 0

Related Questions