• Non ci sono risultati.

MODELLI ARIMA IN SAS ESEMPIO: DECESSI NEL COMUNE DI GENOVA

N/A
N/A
Protected

Academic year: 2021

Condividi "MODELLI ARIMA IN SAS ESEMPIO: DECESSI NEL COMUNE DI GENOVA"

Copied!
12
0
0

Testo completo

(1)

MODELLI ARIMA IN SAS

ESEMPIO: DECESSI NEL COMUNE DI GENOVA

proc arima data=a.tot9102;

identify var=decessi nlag=400 outcov=pluto;

run;quit;

goption reset=(all) htitle=1.5 ftext=SWISSB; symbol7 i=needle ;

axis20 order=-0.5 to 0.5 by 0.1 label=(angle=90); axis21 order=-0.1 to 0.1 by 0.1label=(angle=90);

proc gplot data=pluto;

title 'correlation decessi - dati grezzi fino 2002';

plot corr*lag=7/ vaxis=axis20;

plot partcorr*lag=7/ vaxis=axis20;

run;quit; title ' ';

(2)

proc arima data=a.tot9102;

identify var=decessi (365) nlag=400 outcov=pluto;

run;quit;

proc gplot data=pluto;

title 'correlation decessi - dati differenziati di 365 fino 2002';

plot corr*lag=7/ vaxis=axis20;

plot corr*lag=7/ vaxis=axis21;

plot partcorr*lag=7/ vaxis=axis20;

plot partcorr*lag=7/ vaxis=axis21;

run;quit; title ' ';

(3)

The ARIMA Procedure Name of Variable = Decessi

Period(s) of Differencing 365 Mean of Working Series -0.23245 Standard Deviation 7.769272 Number of Observations 4018 Observation(s) eliminated by differencing 365 Autocorrelations

Lag Covariance Correlation -1 9 8 7 6 5 4 3 2 1 0 1 2 3 4 5 6 7 8 9 1 Std Error 0 60.361595 1.00000 | |********************| 0 1 5.189344 0.08597 | .|** | 0.015776 2 6.349836 0.10520 | .|** | 0.015892 3 7.353399 0.12182 | .|** | 0.016064 4 6.428958 0.10651 | .|** | 0.016293 5 4.464818 0.07397 | .|* | 0.016465

(OMISSIS)

(4)

Partial Autocorrelations

Lag Correlation -1 9 8 7 6 5 4 3 2 1 0 1 2 3 4 5 6 7 8 9 1 1 0.08597 | .|** | 2 0.09853 | .|** | 3 0.10702 | .|** | 4 0.08198 | .|** | 5 0.04139 | .|* | 6 0.05128 | .|* | 7 0.03692 | .|* |

(OMISSIS)

Autocorrelation Check for White Noise To Chi- Pr >

Lag Square DF ChiSq ---Autocorrelations--- 6 231.18 6 <.0001 0.086 0.105 0.122 0.107 0.074 0.086 12 313.89 12 <.0001 0.073 0.067 0.071 0.028 0.046 0.053 18 365.47 18 <.0001 0.069 0.045 0.052 0.035 0.035 0.028 24 383.56 24 <.0001 0.030 0.038 0.042 0.006 -0.001 0.016 30 387.63 30 <.0001 0.004 0.013 0.011 0.008 0.005 0.024 36 398.84 36 <.0001 0.008 0.025 0.014 0.033 0.019 0.021 42 404.35 42 <.0001 0.026 -0.007 0.011 -0.001 0.023 0.002 48 419.62 48 <.0001 -0.020 0.007 0.015 -0.018 0.001 -0.052 54 427.89 54 <.0001 -0.018 -0.020 -0.019 -0.008 -0.022 -0.021 60 433.60 60 <.0001 0.003 -0.017 -0.003 -0.024 -0.008 -0.021 66 460.59 66 <.0001 -0.041 0.035 -0.009 0.032 0.030 -0.040 72 471.05 72 <.0001 -0.015 -0.034 -0.030 -0.001 -0.012 0.012 78 475.50 78 <.0001 0.007 -0.020 0.007 -0.003 -0.008 -0.022 84 477.46 84 <.0001 0.000 -0.000 -0.001 -0.004 0.016 0.014 90 485.21 90 <.0001 0.007 0.019 0.003 0.026 -0.028 -0.007 96 500.19 96 <.0001 0.018 -0.015 -0.029 0.048 0.000 0.003 102 504.52 102 <.0001 0.020 -0.023 0.010 0.004 0.002 0.004 108 522.92 108 <.0001 -0.037 -0.020 -0.027 0.039 0.009 -0.020 114 534.56 114 <.0001 0.015 -0.014 0.015 -0.014 -0.038 -0.023

(OMISSIS)

(5)

proc arima data=a.tot9102;

identify var=decessi(365) nlag=400 noprint outcov=pluto;

estimate p=(1) q=(1)(365) ; run;quit;

The ARIMA Procedure

Conditional Least Squares Estimation

Standard Approx

Parameter Estimate Error t Value Pr > |t| Lag MU -0.22041 0.06199 -3.56 0.0004 0 MA1,1 0.84651 0.02451 34.54 <.0001 1 MA2,1 0.74915 0.01108 67.63 <.0001 365 AR1,1 0.92223 0.01786 51.64 <.0001 1 Constant Estimate -0.01714

Variance Estimate 37.68492 Std Error Estimate 6.138805 AIC 25988.96 SBC 26014.15 Number of Residuals 4018

* AIC and SBC do not include log determinant.

Correlations of Parameter Estimates

Parameter MU MA1,1 MA2,1 AR1,1 MU 1.000 -0.001 0.002 -0.002 MA1,1 -0.001 1.000 -0.002 0.940 MA2,1 0.002 -0.002 1.000 -0.010 AR1,1 -0.002 0.940 -0.010 1.000 Autocorrelation Check of Residuals To Chi- Pr >

Lag Square DF ChiSq ---Autocorrelations--- 6 2.49 3 0.4763 -0.014 0.014 0.010 -0.007 -0.003 0.009 12 6.74 9 0.6646 -0.012 0.006 0.004 -0.029 -0.000 -0.004 18 9.66 15 0.8404 0.010 0.009 0.016 0.006 0.013 -0.009 24 13.50 21 0.8900 -0.013 0.027 0.000 0.004 -0.004 -0.001 30 15.84 27 0.9560 -0.013 -0.010 -0.006 -0.005 -0.015 -0.004 36 17.46 33 0.9879 -0.005 0.003 -0.003 0.016 -0.008 0.007 42 21.44 39 0.9899 0.010 -0.005 0.006 -0.019 0.016 0.014 48 31.13 45 0.9422 -0.023 0.006 -0.000 -0.005 0.005 -0.042 54 35.20 51 0.9551 -0.012 -0.021 -0.012 0.006 0.013 -0.009

(OMISSIS)

354 378.18 351 0.1525 0.009 -0.010 0.003 -0.027 0.006 0.006 360 385.53 357 0.1434 0.009 -0.006 0.020 -0.034 0.003 -0.000 366 399.45 363 0.0910 -0.030 -0.026 0.028 0.006 -0.011 0.025 372 404.11 369 0.1006 0.000 -0.017 -0.022 0.012 -0.007 0.009 378 407.80 375 0.1173 -0.010 -0.009 0.001 0.005 -0.023 -0.010 384 417.71 381 0.0945 0.011 0.014 0.012 0.005 0.015 -0.039 390 425.87 387 0.0843 0.005 -0.021 0.022 0.018 0.009 -0.021 396 429.32 393 0.1000 -0.016 -0.007 -0.002 -0.004 -0.008 0.020 Model for variable Decessi

Estimated Mean -0.22041 Period(s) of Differencing 365 Autoregressive Factors

Factor 1: 1 - 0.92223 B**(1) Moving Average Factors Factor 1: 1 - 0.84651 B**(1) Factor 2: 1 - 0.74915 B**(365)

(6)

ALTRI MODELLI

proc arima data=a.tot9102;

identify var=decessi(365) nlag=400 noprint outcov=pluto;

estimate p=(1) q=(1)(364,365,366) ; run;quit;

The ARIMA Procedure

Conditional Least Squares Estimation

Standard Approx

Parameter Estimate Error t Value Pr > |t| Lag MU -0.22051 0.06600 -3.34 0.0008 0 MA1,1 0.84684 0.02430 34.86 <.0001 1 MA2,1 -0.0003287 0.01110 -0.03 0.9764 364 MA2,2 0.74930 0.01109 67.56 <.0001 365 MA2,3 -0.02274 0.01110 -2.05 0.0406 366 AR1,1 0.92286 0.01765 52.29 <.0001 1

proc arima data=a.tot9102;

identify var=decessi(365) nlag=400 noprint outcov=pluto;

estimate p=(1,2) q=(1)(365) ; run;quit;

The ARIMA Procedure

Conditional Least Squares Estimation

Standard Approx

Parameter Estimate Error t Value Pr > |t| Lag MU -0.22039 0.06108 -3.61 0.0003 0 MA1,1 0.82904 0.03029 27.37 <.0001 1 MA2,1 0.74867 0.01109 67.53 <.0001 365 AR1,1 0.89121 0.03400 26.21 <.0001 1 AR1,2 0.02073 0.01861 1.11 0.2653 2

proc arima data=a.tot9102;

identify var=decessi(365) nlag=400 noprint outcov=pluto;

estimate p=(1) q=(1,2)(365) ; run;quit;

The ARIMA Procedure

Conditional Least Squares Estimation

Standard Approx

Parameter Estimate Error t Value Pr > |t| Lag MU -0.22039 0.06108 -3.61 0.0003 0 MA1,1 0.85141 0.02560 33.26 <.0001 1 MA1,2 -0.01857 0.01713 -1.08 0.2784 2 MA2,1 0.74868 0.01109 67.53 <.0001 365 AR1,1 0.91390 0.02028 45.06 <.0001 1

(7)

FORECAST

proc arima data=a.tot9102;

identify var=decessi(365) nlag=400 noprint outcov=pluto;

estimate p=(1) q=(1)(365) ;

forecast lead=365 out=prev_d_t id=g interval=day noprint;

run;quit;

data prev_d_t2; merge prev_d_t a.totale; by g;run;

axis9 label=none;

axis10 order=('01jan2003'd to '01jan2004'd by quarter) label=none MAJOR=(HEIGHT=.1);

axis12 order=('01jul2003'd to '01sep2003'd by week) ; /*....*/

symbol1 c=red i=j l=1 v=dot h=0.5; symbol2 c=blue i=j l=1 v=dot h=0.5;

symbol3 c=green i=j l=1 v=none; symbol4 c=blue i=j l=1 v=none;

title 'decessi - forecast';

proc gplot data=prev_d_t2;

plot decessi*g=1 /*forecast*g=2*/ l95*g=3 u95*g=4/overlay haxis=axis10;

plot decessi*g=1 /*forecast*g=2*/ l95*g=3 u95*g=4/overlay haxis=axis12;

/*....*/

run;quit;

(8)

ESEMPIO: SCARTO ASSOLUTO DALLA MEDIA DELLE TEMPERATURE

proc arima data=a.tot9102;

identify var= t_min_scartoa nlag=400 outcov=pippo;

run;quit;

proc gplot data= pippo;

title 'correlation temperature - dati grezzi fino 2002';

plot corr*lag=7/ vaxis=axis20;

plot partcorr*lag=7/ vaxis=axis20;

run;quit; title ' ';

(9)

title 'scarto assoluto dalla media delle temperature minime';

proc arima data=a.tot9102;

identify var=t_min_scartoa(1,365) nlag=400 noprint outcov=pippo ; estimate p=(1,2,3) q=(2)(365) ;

forecast lead=365 out=prev_t id=g noprint interval=day;

run;quit;

The ARIMA Procedure

Name of Variable = t_min_scartoa Mean of Working Series 5.16058 Standard Deviation 3.110429 Number of Observations 4383 Autocorrelation Check for White Noise To Chi- Pr >

Lag Square DF ChiSq ---Autocorrelations--- 6 9999.99 6 <.0001 0.833 0.742 0.678 0.635 0.597 0.571 12 9999.99 12 <.0001 0.543 0.524 0.505 0.492 0.477 0.462 18 9999.99 18 <.0001 0.443 0.428 0.418 0.404 0.385 0.367 24 9999.99 24 <.0001 0.349 0.340 0.332 0.321 0.309 0.303 30 9999.99 30 <.0001 0.286 0.271 0.260 0.241 0.226 0.209 36 9999.99 36 <.0001 0.190 0.176 0.157 0.140 0.122 0.109 42 9999.99 42 <.0001 0.096 0.084 0.073 0.055 0.035 0.018 48 9999.99 48 <.0001 0.006 -0.004 -0.014 -0.026 -0.040 -0.051 54 9999.99 54 <.0001 -0.059 -0.069 -0.084 -0.093 -0.109 -0.125 60 9999.99 60 <.0001 -0.139 -0.156 -0.175 -0.193 -0.208 -0.221 Conditional Least Squares Estimation

Standard Approx

Parameter Estimate Error t Value Pr > |t| Lag MU -0.0006338 0.0003855 -1.64 0.1002 0 MA1,1 0.98049 0.0046012 213.10 <.0001 2 MA2,1 0.72378 0.01159 62.44 <.0001 365 AR1,1 -0.34986 0.01573 -22.24 <.0001 1 AR1,2 0.72552 0.01362 53.26 <.0001 2 AR1,3 0.08746 0.01599 5.47 <.0001 3 Constant Estimate -0.00034

Variance Estimate 3.600917 Std Error Estimate 1.897608 AIC 16552.28 SBC 16590.07 Number of Residuals 4017 * AIC and SBC do not include log determinant.

Correlations of Parameter Estimates

Parameter MU MA1,1 MA2,1 AR1,1 AR1,2 AR1,3 MU 1.000 -0.019 -0.035 -0.000 -0.011 -0.005 MA1,1 -0.019 1.000 -0.005 0.010 0.456 0.173 MA2,1 -0.035 -0.005 1.000 0.004 0.022 0.018 AR1,1 -0.000 0.010 0.004 1.000 0.338 -0.687 AR1,2 -0.011 0.456 0.022 0.338 1.000 0.408 AR1,3 -0.005 0.173 0.018 -0.687 0.408 1.000

(10)

Autocorrelation Check of Residuals To Chi- Pr >

Lag Square DF ChiSq ---Autocorrelations--- 6 12.04 1 0.0005 -0.003 -0.028 -0.016 0.026 0.006 0.035 12 23.94 7 0.0012 0.013 0.015 0.010 0.026 0.035 0.024 18 30.39 13 0.0041 -0.005 -0.008 0.024 0.020 0.005 -0.023 24 36.83 19 0.0083 -0.022 -0.017 0.001 -0.005 -0.022 0.018 30 42.50 25 0.0159 -0.012 -0.015 0.026 -0.020 -0.002 0.001 36 47.33 31 0.0305 -0.014 0.000 -0.028 -0.004 -0.012 -0.008 42 57.62 37 0.0165 -0.007 -0.012 0.017 -0.005 -0.028 -0.036 48 61.89 43 0.0309 -0.010 -0.005 -0.004 -0.007 -0.022 -0.020 54 63.15 49 0.0842 0.008 0.002 -0.006 0.013 0.006 -0.001 60 68.93 55 0.0982 0.018 0.003 -0.022 -0.023 -0.004 -0.008 66 84.28 61 0.0259 -0.009 -0.031 0.026 -0.001 0.043 -0.014 72 88.41 67 0.0410 -0.019 0.010 -0.003 -0.007 0.019 0.010 78 95.09 73 0.0423 0.015 0.014 0.001 0.024 -0.015 0.020 84 99.70 79 0.0578 -0.011 -0.004 0.013 -0.013 0.024 0.008 90 109.25 85 0.0394 -0.019 0.021 0.027 0.024 0.002 -0.014 96 116.44 91 0.0373 -0.025 0.017 0.004 -0.020 0.019 0.006 102 124.05 97 0.0334 -0.006 0.003 0.016 -0.007 -0.032 0.022 108 124.53 103 0.0732 -0.003 -0.004 -0.005 0.002 0.008 -0.000 114 136.96 109 0.0362 0.010 0.007 -0.044 -0.026 0.016 -0.004 120 138.02 115 0.0707 -0.002 -0.007 -0.013 0.004 0.000 -0.004 126 144.24 121 0.0735 -0.017 -0.026 0.011 -0.014 -0.011 -0.011 132 149.00 127 0.0886 0.019 -0.017 0.010 -0.009 0.002 0.018 138 151.03 133 0.1357 0.002 -0.000 0.013 0.003 -0.010 0.015 144 153.16 139 0.1944 0.005 -0.008 -0.003 0.012 0.013 0.010 150 159.19 145 0.1986 -0.025 0.007 -0.008 0.026 -0.002 0.006

Model for variable t_min_scartoa Estimated Mean -0.00063 Period(s) of Differencing 1,365

Autoregressive Factors

Factor 1: 1 + 0.34986 B**(1) - 0.72552 B**(2) - 0.08746 B**(3)

Moving Average Factors Factor 1: 1 - 0.98049 B**(2) Factor 2: 1 - 0.72378 B**(365)

(11)

ESEMPIO: DECESSI CON INPUT

SCARTO ASSOLUTO DALLA MEDIA DELLE TEMPERATURE MINIME

/*CROSSCORRELAZIONI CON DATI PREWHINTED*/

title 'decessi con input scarto assoluto dalla media delle temperature minime';

proc arima data=a.tot9102;

identify var=t_min_scartoa(1,365) nlag=400 noprint outcov=pippo ; estimate p=(1,2,3) q=(2)(365) ;

identify var=decessi(365) nlag=400 crosscorr=t_min_scartoa(365) noprint outcov=pluto;

estimate p=(1) q=(1)(365) input=((2)t_min_scartoa);

forecast lead=365 out=prev_d_t id=g interval=day noprint;

run;quit;

data prev_d_t2; merge prev_d_t a.totale; by g;run;

proc gplot data=pluto;

title 'Crosscorrelation decessi / temperature minime scarto - dati prewhinted ';

where crossvar ne ' ';

plot corr*lag=7/ vaxis=axis21;

run;quit; title ' ';

title 'decessi con input scarto ass. dalla media delle temp minime';

proc gplot data=prev_d_t2;

plot decessi*g=1 /*forecast*g=2*/ l95*g=3 u95*g=4/overlay haxis=axis12;

/*....*/

run;quit;

(12)

Riferimenti

Documenti correlati

163/2006 PER L’AFFIDAMENTO DEI SERVIZI TIPOGRAFICI DI INDIRE ED ERASMUS+ PER LA DURATA DI 4 (QUATTRO) ANNI [CIG ACCORDO QUADRO: 6244300F5B] PER LA FORNITURA DI MATERIALE

Acquisire informazioni dettagliate e operative sulle principali attività amministrative e di controllo di una società che offre servizi di elaborazione dati contabili

Si vuole stabilire se esiste una dipendenza fra il flusso di un corso d’acqua (cioè la quantità di acqua che passa in un minuto) e la profondità del corso d’acqua.. Per ottenere

Come sempre, nella costruzione di un modello si cerca un compromesso tra l’aderenza del modello alla realt` a fisica e la semplicit` a del modello stesso: pi` u i compartimenti

(2) Per x negativo 1/x assume valori negativi: valori negativi quanto si voglia piccoli in valore assoluto per x negativo abbastanza grande in valore assoluto (1/x ha limite 0 − per

[r]

Le tavole contengono in prevalenza dati assoluti; sono stati anche calcolati alcuni indicatori demografici (ad esempio, a livello regionale, la speranza di vita a O

In the simplest case of AR(1) models, we have proved the stationarity (with suitable variance of the initial condition) when the parameter satis…es j j &lt; 1.. In general, there