Time series arima forecasting constant line -
Time series arima forecasting constant line -
i have weekly info feb 2013 jan 2015 (104 weeks). tried using auto.arima find best fit.it came out arima(0,1,1). when forecasted next 4 points, produced same values points. pacf plots shows p should @ to the lowest degree 1. , acf graph gradually trending down, not closing zero. can explain me why produce constant line?
coefficients: ma1 -0.3558 s.e. 0.1009
sigma^2 estimated 579288: log likelihood=-829.6 aic=1663.2 aicc=1663.32 bic=1668.47
below info
week orders 201301 2974 201302 3221.3 201303 3295 201304 3750.5 201305 3542.8 201306 3238.3 201307 3246.3 201308 3265.5 201309 3551.6 201310 4019.4 201311 3369.7 201312 3675.1 201313 3717.3 201314 3621 201315 3607 201316 3752.4 201317 3834 201318 4218.7 201319 3850.8 201320 3881.2 201321 3825.2 201322 3398.2 201323 4104.1 201324 3769.7 201325 3700.5 201326 3839.6 201327 3902.9 201328 3948.3 201329 4319.8 201330 4034.7 201331 3952.9 201332 3817.3 201333 4278.4 201334 5083.5 201335 5776.1 201336 5815.8 201337 6085.5 201338 5883.4 201339 6132.9 201340 6257.2 201341 5719.7 201342 5939.2 201343 4292 201344 5154.2 201345 4899.2 201346 4931.7 201347 4154.7 201348 4974.7 201349 5123 201350 5194.5 201351 4673.3 201352 4750.5 201401 4921.8 201402 4972.6 201403 5153.8 201404 5107.7 201405 5013.2 201406 5396.40000000001 201407 5213.4 201408 4985.3 201409 4771.3 201410 4632.4 201411 4815.5 201412 4742.3 201413 3921.3 201414 4316.6 201415 5088.3 201416 4764.5 201417 5144.5 201418 5531.9 201419 5378.2 201420 5547.1 201421 5096.2 201422 5528.1 201423 6564.9 201424 4797.7 201425 9488.8 201426 8955.7 201427 8680.3 201428 7821.1 201429 8692.3 201430 5595.2 201431 5468.5 201432 5562.6 201433 5930 201434 7011 201435 7995 201436 7099 201437 7056 201438 6483 201439 6697.5 201440 7802.5 201441 8038 201442 7972 201443 5816 201444 7150.5 201445 6467.5 201446 6181.5 201447 4735 201448 6187 201449 6725 201450 5773.5 201451 5558.5 201452 5369.76 201501 5585.58 201502 5168.28 201503 6505.84 201504 5562.24
time-series
Comments
Post a Comment