데이터분석

[23.07.04] 데이터 시각화(시계열 분석) - 23(1)

gmwoo 2023. 7. 4. 16:48

시계열 분석¶

연도별 계절별 월별 일별 시 분 초별로 시간의 흐름에 따라 관측된 자료

시계열 자료의 여러 형태

시계열 그림 시간의 경과에 따라 시계열자료의 값이 변하는 것을 그린 그림

목적

1) 미래 예측

2) 시스템 또는 확률과정의 이해와 제어

1. Numpy의 profit으로 회귀(regression) 분석하기¶

1) 야후 파이낸스 데이터를 가져오기 위한 yfinance
```
 * pip install yfinance
```
2) 예측 모델링 및 시각화를 위한 prophet
```
* pip install prophet
```
3) Prophets는 Plotly 기반으로 데이터를 시각화하므로
```
* pip install plotly
```

In [1]:

import warnings
warnings.filterwarnings('ignore')
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from prophet import Prophet

In [2]:

import matplotlib.pyplot as plt
from matplotlib import font_manager, rc
plt.rcParams['axes.unicode_minus'] = False   # - 기호 깨지는 것 방지
# f_path = "/Library/Fonts/AppleGothic.ttf"   -> MAC
f_path = "C:/Windows/Fonts/malgun.ttf"
font_name = font_manager.FontProperties(fname=f_path).get_name()
rc('font', family=font_name)

홈페이지 방문객 숫자(날짜, 방문객 수)

In [3]:

pweb = pd.read_csv('../../data/08. PinkWink Web Traffic.csv',
                  encoding='utf-8', thousands=',',
                  names=['date', 'hit'], index_col=0)
pweb = pweb[pweb['hit'].notnull()]
pweb.head()

Out[3]:

	hit
date
16. 7. 1.	766.0
16. 7. 2.	377.0
16. 7. 3.	427.0
16. 7. 4.	902.0
16. 7. 5.	850.0

In [4]:

pweb['hit'].plot(figsize=(12,4), grid=True);

In [5]:

time = np.arange(0,len(pweb)) # 시간(time) 축 생성 (0~364)
# print(time)
traffic = pweb['hit'].values # 웹 트래픽 traffic 변수에 저장
fx = np.linspace(0, time[-1], 1000)

 [  0.           0.36436436   0.72872873   1.09309309   1.45745746
   1.82182182   2.18618619   2.55055055   2.91491491   3.27927928
   3.64364364   4.00800801   4.37237237   4.73673674   5.1011011
   5.46546547   5.82982983   6.19419419   6.55855856   6.92292292
   7.28728729   7.65165165   8.01601602   8.38038038   8.74474474
   9.10910911   9.47347347   9.83783784  10.2022022   10.56656657
  10.93093093  11.2952953   11.65965966  12.02402402  12.38838839
  12.75275275  13.11711712  13.48148148  13.84584585  14.21021021
  14.57457457  14.93893894  15.3033033   15.66766767  16.03203203
  16.3963964   16.76076076  17.12512513  17.48948949  17.85385385
  18.21821822  18.58258258  18.94694695  19.31131131  19.67567568
  20.04004004  20.4044044   20.76876877  21.13313313  21.4974975
  21.86186186  22.22622623  22.59059059  22.95495495  23.31931932
  23.68368368  24.04804805  24.41241241  24.77677678  25.14114114
  25.50550551  25.86986987  26.23423423  26.5985986   26.96296296
  27.32732733  27.69169169  28.05605606  28.42042042  28.78478478
  29.14914915  29.51351351  29.87787788  30.24224224  30.60660661
  30.97097097  31.33533534  31.6996997   32.06406406  32.42842843
  32.79279279  33.15715716  33.52152152  33.88588589  34.25025025
  34.61461461  34.97897898  35.34334334  35.70770771  36.07207207
  36.43643644  36.8008008   37.16516517  37.52952953  37.89389389
  38.25825826  38.62262262  38.98698699  39.35135135  39.71571572
  40.08008008  40.44444444  40.80880881  41.17317317  41.53753754
  41.9019019   42.26626627  42.63063063  42.99499499  43.35935936
  43.72372372  44.08808809  44.45245245  44.81681682  45.18118118
  45.54554555  45.90990991  46.27427427  46.63863864  47.003003
  47.36736737  47.73173173  48.0960961   48.46046046  48.82482482
  49.18918919  49.55355355  49.91791792  50.28228228  50.64664665
  51.01101101  51.37537538  51.73973974  52.1041041   52.46846847
  52.83283283  53.1971972   53.56156156  53.92592593  54.29029029
  54.65465465  55.01901902  55.38338338  55.74774775  56.11211211
  56.47647648  56.84084084  57.20520521  57.56956957  57.93393393
  58.2982983   58.66266266  59.02702703  59.39139139  59.75575576
  60.12012012  60.48448448  60.84884885  61.21321321  61.57757758
  61.94194194  62.30630631  62.67067067  63.03503504  63.3993994
  63.76376376  64.12812813  64.49249249  64.85685686  65.22122122
  65.58558559  65.94994995  66.31431431  66.67867868  67.04304304
  67.40740741  67.77177177  68.13613614  68.5005005   68.86486486
  69.22922923  69.59359359  69.95795796  70.32232232  70.68668669
  71.05105105  71.41541542  71.77977978  72.14414414  72.50850851
  72.87287287  73.23723724  73.6016016   73.96596597  74.33033033
  74.69469469  75.05905906  75.42342342  75.78778779  76.15215215
  76.51651652  76.88088088  77.24524525  77.60960961  77.97397397
  78.33833834  78.7027027   79.06706707  79.43143143  79.7957958
  80.16016016  80.52452452  80.88888889  81.25325325  81.61761762
  81.98198198  82.34634635  82.71071071  83.07507508  83.43943944
  83.8038038   84.16816817  84.53253253  84.8968969   85.26126126
  85.62562563  85.98998999  86.35435435  86.71871872  87.08308308
  87.44744745  87.81181181  88.17617618  88.54054054  88.9049049
  89.26926927  89.63363363  89.997998    90.36236236  90.72672673
  91.09109109  91.45545546  91.81981982  92.18418418  92.54854855
  92.91291291  93.27727728  93.64164164  94.00600601  94.37037037
  94.73473473  95.0990991   95.46346346  95.82782783  96.19219219
  96.55655656  96.92092092  97.28528529  97.64964965  98.01401401
  98.37837838  98.74274274  99.10710711  99.47147147  99.83583584
 100.2002002  100.56456456 100.92892893 101.29329329 101.65765766
 102.02202202 102.38638639 102.75075075 103.11511512 103.47947948
 103.84384384 104.20820821 104.57257257 104.93693694 105.3013013
 105.66566567 106.03003003 106.39439439 106.75875876 107.12312312
 107.48748749 107.85185185 108.21621622 108.58058058 108.94494494
 109.30930931 109.67367367 110.03803804 110.4024024  110.76676677
 111.13113113 111.4954955  111.85985986 112.22422422 112.58858859
 112.95295295 113.31731732 113.68168168 114.04604605 114.41041041
 114.77477477 115.13913914 115.5035035  115.86786787 116.23223223
 116.5965966  116.96096096 117.32532533 117.68968969 118.05405405
 118.41841842 118.78278278 119.14714715 119.51151151 119.87587588
 120.24024024 120.6046046  120.96896897 121.33333333 121.6976977
 122.06206206 122.42642643 122.79079079 123.15515516 123.51951952
 123.88388388 124.24824825 124.61261261 124.97697698 125.34134134
 125.70570571 126.07007007 126.43443443 126.7987988  127.16316316
 127.52752753 127.89189189 128.25625626 128.62062062 128.98498498
 129.34934935 129.71371371 130.07807808 130.44244244 130.80680681
 131.17117117 131.53553554 131.8998999  132.26426426 132.62862863
 132.99299299 133.35735736 133.72172172 134.08608609 134.45045045
 134.81481481 135.17917918 135.54354354 135.90790791 136.27227227
 136.63663664 137.001001   137.36536537 137.72972973 138.09409409
 138.45845846 138.82282282 139.18718719 139.55155155 139.91591592
 140.28028028 140.64464464 141.00900901 141.37337337 141.73773774
 142.1021021  142.46646647 142.83083083 143.1951952  143.55955956
 143.92392392 144.28828829 144.65265265 145.01701702 145.38138138
 145.74574575 146.11011011 146.47447447 146.83883884 147.2032032
 147.56756757 147.93193193 148.2962963  148.66066066 149.02502503
 149.38938939 149.75375375 150.11811812 150.48248248 150.84684685
 151.21121121 151.57557558 151.93993994 152.3043043  152.66866867
 153.03303303 153.3973974  153.76176176 154.12612613 154.49049049
 154.85485485 155.21921922 155.58358358 155.94794795 156.31231231
 156.67667668 157.04104104 157.40540541 157.76976977 158.13413413
 158.4984985  158.86286286 159.22722723 159.59159159 159.95595596
 160.32032032 160.68468468 161.04904905 161.41341341 161.77777778
 162.14214214 162.50650651 162.87087087 163.23523524 163.5995996
 163.96396396 164.32832833 164.69269269 165.05705706 165.42142142
 165.78578579 166.15015015 166.51451451 166.87887888 167.24324324
 167.60760761 167.97197197 168.33633634 168.7007007  169.06506507
 169.42942943 169.79379379 170.15815816 170.52252252 170.88688689
 171.25125125 171.61561562 171.97997998 172.34434434 172.70870871
 173.07307307 173.43743744 173.8018018  174.16616617 174.53053053
 174.89489489 175.25925926 175.62362362 175.98798799 176.35235235
 176.71671672 177.08108108 177.44544545 177.80980981 178.17417417
 178.53853854 178.9029029  179.26726727 179.63163163 179.995996
 180.36036036 180.72472472 181.08908909 181.45345345 181.81781782
 182.18218218 182.54654655 182.91091091 183.27527528 183.63963964
 184.004004   184.36836837 184.73273273 185.0970971  185.46146146
 185.82582583 186.19019019 186.55455455 186.91891892 187.28328328
 187.64764765 188.01201201 188.37637638 188.74074074 189.10510511
 189.46946947 189.83383383 190.1981982  190.56256256 190.92692693
 191.29129129 191.65565566 192.02002002 192.38438438 192.74874875
 193.11311311 193.47747748 193.84184184 194.20620621 194.57057057
 194.93493493 195.2992993  195.66366366 196.02802803 196.39239239
 196.75675676 197.12112112 197.48548549 197.84984985 198.21421421
 198.57857858 198.94294294 199.30730731 199.67167167 200.03603604
 200.4004004  200.76476476 201.12912913 201.49349349 201.85785786
 202.22222222 202.58658659 202.95095095 203.31531532 203.67967968
 204.04404404 204.40840841 204.77277277 205.13713714 205.5015015
 205.86586587 206.23023023 206.59459459 206.95895896 207.32332332
 207.68768769 208.05205205 208.41641642 208.78078078 209.14514515
 209.50950951 209.87387387 210.23823824 210.6026026  210.96696697
 211.33133133 211.6956957  212.06006006 212.42442442 212.78878879
 213.15315315 213.51751752 213.88188188 214.24624625 214.61061061
 214.97497497 215.33933934 215.7037037  216.06806807 216.43243243
 216.7967968  217.16116116 217.52552553 217.88988989 218.25425425
 218.61861862 218.98298298 219.34734735 219.71171171 220.07607608
 220.44044044 220.8048048  221.16916917 221.53353353 221.8978979
 222.26226226 222.62662663 222.99099099 223.35535536 223.71971972
 224.08408408 224.44844845 224.81281281 225.17717718 225.54154154
 225.90590591 226.27027027 226.63463463 226.998999   227.36336336
 227.72772773 228.09209209 228.45645646 228.82082082 229.18518519
 229.54954955 229.91391391 230.27827828 230.64264264 231.00700701
 231.37137137 231.73573574 232.1001001  232.46446446 232.82882883
 233.19319319 233.55755756 233.92192192 234.28628629 234.65065065
 235.01501502 235.37937938 235.74374374 236.10810811 236.47247247
 236.83683684 237.2012012  237.56556557 237.92992993 238.29429429
 238.65865866 239.02302302 239.38738739 239.75175175 240.11611612
 240.48048048 240.84484484 241.20920921 241.57357357 241.93793794
 242.3023023  242.66666667 243.03103103 243.3953954  243.75975976
 244.12412412 244.48848849 244.85285285 245.21721722 245.58158158
 245.94594595 246.31031031 246.67467467 247.03903904 247.4034034
 247.76776777 248.13213213 248.4964965  248.86086086 249.22522523
 249.58958959 249.95395395 250.31831832 250.68268268 251.04704705
 251.41141141 251.77577578 252.14014014 252.5045045  252.86886887
 253.23323323 253.5975976  253.96196196 254.32632633 254.69069069
 255.05505506 255.41941942 255.78378378 256.14814815 256.51251251
 256.87687688 257.24124124 257.60560561 257.96996997 258.33433433
 258.6986987  259.06306306 259.42742743 259.79179179 260.15615616
 260.52052052 260.88488488 261.24924925 261.61361361 261.97797798
 262.34234234 262.70670671 263.07107107 263.43543544 263.7997998
 264.16416416 264.52852853 264.89289289 265.25725726 265.62162162
 265.98598599 266.35035035 266.71471471 267.07907908 267.44344344
 267.80780781 268.17217217 268.53653654 268.9009009  269.26526527
 269.62962963 269.99399399 270.35835836 270.72272272 271.08708709
 271.45145145 271.81581582 272.18018018 272.54454454 272.90890891
 273.27327327 273.63763764 274.002002   274.36636637 274.73073073
 275.0950951  275.45945946 275.82382382 276.18818819 276.55255255
 276.91691692 277.28128128 277.64564565 278.01001001 278.37437437
 278.73873874 279.1031031  279.46746747 279.83183183 280.1961962
 280.56056056 280.92492492 281.28928929 281.65365365 282.01801802
 282.38238238 282.74674675 283.11111111 283.47547548 283.83983984
 284.2042042  284.56856857 284.93293293 285.2972973  285.66166166
 286.02602603 286.39039039 286.75475475 287.11911912 287.48348348
 287.84784785 288.21221221 288.57657658 288.94094094 289.30530531
 289.66966967 290.03403403 290.3983984  290.76276276 291.12712713
 291.49149149 291.85585586 292.22022022 292.58458458 292.94894895
 293.31331331 293.67767768 294.04204204 294.40640641 294.77077077
 295.13513514 295.4994995  295.86386386 296.22822823 296.59259259
 296.95695696 297.32132132 297.68568569 298.05005005 298.41441441
 298.77877878 299.14314314 299.50750751 299.87187187 300.23623624
 300.6006006  300.96496496 301.32932933 301.69369369 302.05805806
 302.42242242 302.78678679 303.15115115 303.51551552 303.87987988
 304.24424424 304.60860861 304.97297297 305.33733734 305.7017017
 306.06606607 306.43043043 306.79479479 307.15915916 307.52352352
 307.88788789 308.25225225 308.61661662 308.98098098 309.34534535
 309.70970971 310.07407407 310.43843844 310.8028028  311.16716717
 311.53153153 311.8958959  312.26026026 312.62462462 312.98898899
 313.35335335 313.71771772 314.08208208 314.44644645 314.81081081
 315.17517518 315.53953954 315.9039039  316.26826827 316.63263263
 316.996997   317.36136136 317.72572573 318.09009009 318.45445445
 318.81881882 319.18318318 319.54754755 319.91191191 320.27627628
 320.64064064 321.00500501 321.36936937 321.73373373 322.0980981
 322.46246246 322.82682683 323.19119119 323.55555556 323.91991992
 324.28428428 324.64864865 325.01301301 325.37737738 325.74174174
 326.10610611 326.47047047 326.83483483 327.1991992  327.56356356
 327.92792793 328.29229229 328.65665666 329.02102102 329.38538539
 329.74974975 330.11411411 330.47847848 330.84284284 331.20720721
 331.57157157 331.93593594 332.3003003  332.66466466 333.02902903
 333.39339339 333.75775776 334.12212212 334.48648649 334.85085085
 335.21521522 335.57957958 335.94394394 336.30830831 336.67267267
 337.03703704 337.4014014  337.76576577 338.13013013 338.49449449
 338.85885886 339.22322322 339.58758759 339.95195195 340.31631632
 340.68068068 341.04504505 341.40940941 341.77377377 342.13813814
 342.5025025  342.86686687 343.23123123 343.5955956  343.95995996
 344.32432432 344.68868869 345.05305305 345.41741742 345.78178178
 346.14614615 346.51051051 346.87487487 347.23923924 347.6036036
 347.96796797 348.33233233 348.6966967  349.06106106 349.42542543
 349.78978979 350.15415415 350.51851852 350.88288288 351.24724725
 351.61161161 351.97597598 352.34034034 352.7047047  353.06906907
 353.43343343 353.7977978  354.16216216 354.52652653 354.89089089
 355.25525526 355.61961962 355.98398398 356.34834835 356.71271271
 357.07707708 357.44144144 357.80580581 358.17017017 358.53453453
 358.8988989  359.26326326 359.62762763 359.99199199 360.35635636
 360.72072072 361.08508509 361.44944945 361.81381381 362.17817818
 362.54254254 362.90690691 363.27127127 363.63563564 364.        ]

평균제곱오차(mean squared error, mse)
오차 = (예측값 - 실제값) ** 2

In [6]:

def error(f, x, y):
    return np.sqrt(np.mean((f(x)-y)**2))

In [7]:

# polyfit(x, y, n차)
fp1 = np.polyfit(time, traffic, 1)  # 다항식의 입력값 x, y, 1차
f1 = np.poly1d(fp1)
fp2 = np.polyfit(time, traffic, 2)  # 다항식의 입력값 x, y, 2차
f2 = np.poly1d(fp2)
fp3 = np.polyfit(time, traffic, 3)  # 다항식의 입력값 x, y, 3차
f3 = np.poly1d(fp3)
fp15 = np.polyfit(time, traffic, 15) # 다항식의 입력값 x, y, 15차
f15 = np.poly1d(fp15)
print(error(f1, time, traffic))
print(error(f2, time, traffic))
print(error(f3, time, traffic))
print(error(f15, time, traffic))

 430.85973081109626
430.6284101894695
429.53280466762925
330.4777304274343

In [8]:

plt.figure(figsize=(10,6))
plt.scatter(time, traffic, s=10)
plt.plot(fx, f1(fx), lw=4, label='f1')
plt.plot(fx, f2(fx), lw=4, label='f2')
plt.plot(fx, f3(fx), lw=4, label='f3')
plt.plot(fx, f15(fx), lw=4, label='f15')  # overfitting
plt.grid(True, linestyle='-', color='0.75')
plt.legend(loc=2)

2. Prophet 모듈을 이용한 forecast 예측¶

In [9]:

df = pd.DataFrame({'ds':pweb.index, 'y':pweb['hit']})
df.reset_index(inplace=True)
df['ds'] = pd.to_datetime(df['ds'], format="%y. %m. %d.")

Out[9]:

	date	ds	y
0	16. 7. 1.	2016-07-01	766.0
1	16. 7. 2.	2016-07-02	377.0
2	16. 7. 3.	2016-07-03	427.0
3	16. 7. 4.	2016-07-04	902.0
4	16. 7. 5.	2016-07-05	850.0

In [10]:

del df['date'] # df 변수를 생성했으므로 삭제
# Prophet() 함수를 사용 시
# 주기성이 연단위(yearly_seasonality) 및 일단위(daily_seasonality=True)로 있다고 알려줌
# Prophet 클래스 객체를 만들고 시계열 데이터를 입력으로 fit 메서드를 호출
m = Prophet(yearly_seasonality=True, daily_seasonality=True)

 16:45:53 - cmdstanpy - INFO - Chain [1] start processing
16:45:53 - cmdstanpy - INFO - Chain [1] done processing

Out[10]:

<prophet.forecaster.Prophet at 0x1b874521d50>

In [11]:

# make_future_dataframe : 예측 날짜 구간 생성()
# 16. 7. 1. ~ 17. 6. 30. 홈페이지 방문자 수
future = m.make_future_dataframe(periods=60)
future.head()

Out[11]:

	ds
0	2016-07-01
1	2016-07-02
2	2016-07-03
3	2016-07-04
4	2016-07-05

predict: 신뢰구간을 포함한 예측 실행

In [12]:

forecast = m.predict(future)
forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']].tail()

Out[12]:

	ds	yhat	yhat_lower	yhat_upper
420	2017-08-25	901.398379	703.775237	1110.598991
421	2017-08-26	486.266408	260.578665	691.996784
422	2017-08-27	618.928059	414.174886	825.380619
423	2017-08-28	1171.734066	963.785863	1375.587713
424	2017-08-29	1207.743873	987.267201	1429.176432

In [13]:

# plot: 원래의 시계열 데이터와 예측 데이터
m.plot(forecast)

Out[13]:

In [14]:

# plot_components: 선형회귀 및 계절성 성분별로 분리
m.plot_components(forecast)

Out[14]:

3. Seasonal 시계열 분석으로 주식 데이터 분석하기¶

In [15]:

# https://finance.yahoo.com/
import datetime as dt
import yfinance as yf
company = 'TATAELXSI.NS'
# Define a start date and End Date
start = dt.datetime(2021, 1, 1)
end = dt.datetime(2023, 5, 1)
# Read Stock Price Data
data = yf.download(company, start, end)
data.tail(10)
# oppen 시가, high 고가, low 저가, close 종가, Adj close 조정종가, volume 거래량

 [*********************100%***********************]  1 of 1 completed

Out[15]:

	Open	High	Low	Close	Adj Close	Volume
Date
2023-04-17	6237.950195	6300.0	6180.000000	6289.750000	6240.825195	108512
2023-04-18	6298.750000	6396.0	6295.000000	6367.299805	6317.771973	100611
2023-04-19	6368.000000	6375.0	6290.049805	6307.200195	6258.139648	69527
2023-04-20	6316.000000	6364.5	6235.350098	6251.649902	6203.021484	90809
2023-04-21	6269.000000	6300.0	6205.549805	6281.000000	6232.143555	66308
2023-04-24	6280.850098	6330.0	6220.049805	6252.950195	6204.312012	61632
2023-04-25	6253.000000	6307.0	6235.299805	6271.200195	6222.419922	47904
2023-04-26	6269.000000	6300.0	6240.000000	6290.850098	6241.916992	56123
2023-04-27	6284.000000	6635.0	6265.149902	6578.200195	6527.031738	493002
2023-04-28	6629.000000	6687.0	6565.649902	6643.600098	6591.922852	191519

In [16]:

start = '2012-7-1'
end = '2022-7-31'
KIA = yf.download('000270.KS', start, end) # 기아차

 [*********************100%***********************]  1 of 1 completed

Out[16]:

	Open	High	Low	Close	Adj Close	Volume
Date
2012-07-02	75600.0	75900.0	75000.0	75300.0	57019.429688	950363
2012-07-03	75800.0	75800.0	73100.0	73500.0	55656.425781	2542262
2012-07-04	74300.0	74500.0	73600.0	74500.0	56413.644531	1786898
2012-07-05	74600.0	74800.0	73800.0	74100.0	56110.750000	835637
2012-07-06	74400.0	74500.0	73500.0	73700.0	55807.871094	758448

In [17]:

start = '2012-7-1'
end = '2022-7-31'
KAKAO = yf.download('035720.KS', start, end) # 카카오
KAKAO.head()

 [*********************100%***********************]  1 of 1 completed

Out[17]:

	Open	High	Low	Close	Adj Close	Volume
Date
2012-07-02	20348.544922	20993.250000	20207.515625	20207.515625	19494.722656	396841
2012-07-03	20267.957031	20650.750000	20227.662109	20550.015625	19825.144531	296579
2012-07-04	20670.898438	20751.486328	20288.103516	20408.986328	19689.089844	351768
2012-07-05	20449.279297	20711.191406	20348.544922	20590.308594	19864.013672	239518
2012-07-06	20529.867188	20691.044922	20429.132812	20691.044922	19961.195312	132778

In [18]:

start = '2012-7-1'
end = '2022-7-31'
SAM = yf.download('005930.KS', start, end) # 삼성

 [*********************100%***********************]  1 of 1 completed

Out[18]:

	Open	High	Low	Close	Adj Close	Volume
Date
2012-07-02	24160.0	24180.0	23420.0	23480.0	18562.669922	19106750
2012-07-03	23640.0	23860.0	23400.0	23500.0	18578.484375	15801550
2012-07-04	23640.0	23940.0	23560.0	23820.0	18831.460938	19158750
2012-07-05	23740.0	23880.0	23540.0	23700.0	18736.601562	9068550
2012-07-06	23880.0	23880.0	23060.0	23220.0	18357.119141	21158900

In [19]:

SAM['Close'].plot(figsize=(12,6), grid=True)

Out[19]:

<Axes: xlabel='Date'>

In [20]:

SAM_trunc = SAM[:'2022-12-31']

Out[20]:

	Open	High	Low	Close	Adj Close	Volume
Date
2012-07-02	24160.0	24180.0	23420.0	23480.0	18562.669922	19106750
2012-07-03	23640.0	23860.0	23400.0	23500.0	18578.484375	15801550
2012-07-04	23640.0	23940.0	23560.0	23820.0	18831.460938	19158750
2012-07-05	23740.0	23880.0	23540.0	23700.0	18736.601562	9068550
2012-07-06	23880.0	23880.0	23060.0	23220.0	18357.119141	21158900
...	...	...	...	...	...	...
2022-07-25	60900.0	61900.0	60800.0	61100.0	59958.777344	9193681
2022-07-26	60800.0	61900.0	60800.0	61700.0	60547.570312	6597211
2022-07-27	61300.0	61900.0	61200.0	61800.0	60645.703125	7320997
2022-07-28	62300.0	62600.0	61600.0	61900.0	60743.835938	10745302
2022-07-29	62400.0	62600.0	61300.0	61400.0	60253.171875	15093120

2479 rows × 6 columns

In [21]:

df = pd.DataFrame({'ds':SAM_trunc.index, 'y':SAM_trunc['Close']})
df.reset_index(inplace=True)
del df['Date']

Out[21]:

	ds	y
0	2012-07-02	23480.0
1	2012-07-03	23500.0
2	2012-07-04	23820.0
3	2012-07-05	23700.0
4	2012-07-06	23220.0

In [22]:

# 주기성이 일단위로 있다고 알려줌
m = Prophet(daily_seasonality=True)

 16:45:59 - cmdstanpy - INFO - Chain [1] start processing
16:46:01 - cmdstanpy - INFO - Chain [1] done processing

Out[22]:

<prophet.forecaster.Prophet at 0x1b876b5ccd0>

In [23]:

# 위의 df에서 1년치를 예측하겠다
future = m.make_future_dataframe(periods=365)
future.tail()

Out[23]:

	ds
2839	2023-07-25
2840	2023-07-26
2841	2023-07-27
2842	2023-07-28
2843	2023-07-29

In [24]:

forecast = m.predict(future)
forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']].tail()

Out[24]:

	ds	yhat	yhat_lower	yhat_upper
2839	2023-07-25	79309.637559	71225.228519	87876.370676
2840	2023-07-26	79304.453351	71186.993432	88040.565048
2841	2023-07-27	79392.731979	71285.285187	87915.863145
2842	2023-07-28	79357.274911	71483.087512	87833.151436
2843	2023-07-29	78919.623670	70468.300319	87604.851549

In [25]:

m.plot(forecast)

Out[25]:

In [26]:

m.plot_components(forecast)

Out[26]:

In [27]:

start = '2014-1-1'
end = '2017-7-31'
SAM = yf.download('005930.KS', start, end) # 삼성
SAM['Close'].plot(figsize=(12,6), grid=True)

 [*********************100%***********************]  1 of 1 completed

Out[27]:

<Axes: xlabel='Date'>

In [28]:

# 일부 데이터를 잘라서 forecast 수행 => 2014-1-1 ~ 2017-7-31
SAM_trunc = SAM[:'2017-05-31']
SAM_trunc['Close'].plot(figsize=(12,6), grid=True)

Out[28]:

<Axes: xlabel='Date'>

In [29]:

df = pd.DataFrame({'ds':SAM_trunc.index, 'y':SAM_trunc['Close']})
df.reset_index(inplace=True)
del df['Date']

In [30]:

m = Prophet(daily_seasonality=True)

 16:46:06 - cmdstanpy - INFO - Chain [1] start processing
16:46:06 - cmdstanpy - INFO - Chain [1] done processing

Out[30]:

<prophet.forecaster.Prophet at 0x1b876b5d6c0>

In [31]:

future = m.make_future_dataframe(periods=61)
future.tail()

Out[31]:

	ds
897	2017-07-27
898	2017-07-28
899	2017-07-29
900	2017-07-30
901	2017-07-31

In [32]:

forecast = m.predict(future)
m.plot(forecast);

In [33]:

plt.figure(figsize=(12,6))
plt.plot(SAM.index, SAM['Close'], label='real')
plt.plot(forecast['ds'], forecast['yhat'], label='forecast')
plt.legend()

4. Growth Model¶

주기성을 띠면서 점점 성장하는 모델

In [34]:

df = pd.read_csv('../../data/08. example_wp_R.csv')
df['y'] = np.log(df['y'])   # 로그 변환

Out[34]:

	ds	y
0	2008-01-30	5.976351
1	2008-01-16	6.049733
2	2008-01-17	6.011267
3	2008-01-14	5.953243
4	2008-01-15	5.910797
...	...	...
2858	2015-12-11	7.834788
2859	2015-12-12	7.360104
2860	2015-12-13	7.479864
2861	2015-12-18	7.765145
2862	2015-12-19	7.220374

2863 rows × 2 columns

In [35]:

# 예측값의 최대 상한, 하한값을 제어할 수 있음
df['cap'] = 8.5  # 예측값의 상한값
df['floor'] = 6  # 예측값의 하한값

In [36]:

# daily_seasonality=True : 주기성이 일단위라고 알려줌
# growth 파라미터를 설정 : 로지스틱 함수
m = Prophet(growth='logistic', daily_seasonality=True)

 16:46:08 - cmdstanpy - INFO - Chain [1] start processing
16:46:08 - cmdstanpy - INFO - Chain [1] done processing

Out[36]:

<prophet.forecaster.Prophet at 0x1b878463460>

In [37]:

# Prophet은 Linear한 모델을 사용 - 제어하지 않으면 해당 카테고리의
# 최대 사이즈를 넘을 수도 있음(1825일 -> 5년)
future = m.make_future_dataframe(periods=1826)
future['cap'] = 8.5
future['floor'] = 6
fcst = m.predict(future)
m.plot(fcst);

In [38]:

forecast = m.predict(future)
m.plot_components(forecast);

1. 관심있는 주식을 분석해 보세요.¶

2. 기간 : 10년치¶

3. 6개월 후 예측해 보세요¶

<한화에어로스페이스 주가예측>¶

In [39]:

start = '2012-7-1'
end = '2023-7-31'
hanwhaAero = yf.download('012450.KS', start, end) # 한화에어로스페이스
df_han = pd.DataFrame({'ds':hanwhaAero.index, 'y':hanwhaAero['Close']})
df_han.reset_index(inplace=True)
del df_han['Date']
m = Prophet(daily_seasonality=True)
m.fit(df_han)
future = m.make_future_dataframe(periods=181)
forecast = m.predict(future)
forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']].tail()

 [*********************100%***********************]  1 of 1 completed

 16:46:40 - cmdstanpy - INFO - Chain [1] start processing
16:46:41 - cmdstanpy - INFO - Chain [1] done processing

Out[39]:

	ds	yhat	yhat_lower	yhat_upper
2883	2023-12-28	105717.788834	96303.582508	114361.737294
2884	2023-12-29	105756.688165	97064.901554	114597.808848
2885	2023-12-30	107089.836948	98152.359358	115753.696234
2886	2023-12-31	107187.987911	98441.743063	115587.641210
2887	2024-01-01	106143.506882	97443.648866	115126.941530

In [40]:

plt.figure(figsize=(12,6))
plt.plot(hanwhaAero.index, hanwhaAero['Close'], label='real')
plt.plot(forecast['ds'], forecast['yhat'], label='forecast')
plt.legend()

In [41]:

forecast = m.predict(future)
m.plot(forecast);

저작자표시

'데이터분석' 카테고리의 다른 글

[23.07.06] 인스타그램 크롤링 - 25(1) (1)	2023.07.06
[23.07.05] 데이터 시각화(WordCloud) - 24(1) (0)	2023.07.05
[23.07.03] 데이터 시각화(인구 소멸) - 22(2) (0)	2023.07.03
[23.07.03] 데이터 시각화(기름 제일 싼 곳) - 22(1) (0)	2023.07.03
[23.06.30] 데이터 시각화(따릉이) - 21(3) (0)	2023.06.30

현재글[23.07.04] 데이터 시각화(시계열 분석) - 23(1)

내 블로그 - 관리자 홈 전환	`Q` `Q`
새 글 쓰기	`W` `W`

글 수정 (권한 있는 경우)	`E` `E`
댓글 영역으로 이동	`C` `C`

이 페이지의 URL 복사	`S` `S`
맨 위로 이동	`T` `T`
티스토리 홈 이동	`H` `H`
단축키 안내	`Shift` + `/` `⇧` + `/`

Woogi

[23.07.04] 데이터 시각화(시계열 분석) - 23(1)

시계열 분석¶

1. Numpy의 profit으로 회귀(regression) 분석하기¶

2. Prophet 모듈을 이용한 forecast 예측¶

3. Seasonal 시계열 분석으로 주식 데이터 분석하기¶

4. Growth Model¶

1. 관심있는 주식을 분석해 보세요.¶

2. 기간 : 10년치¶

3. 6개월 후 예측해 보세요¶

<한화에어로스페이스 주가예측>¶

'데이터분석' 카테고리의 다른 글

'데이터분석'의 다른글

티스토리툴바

단축키

내 블로그

블로그 게시글

모든 영역

[23.07.04] 데이터 시각화(시계열 분석) - 23(1)

시계열 분석¶

1. Numpy의 profit으로 회귀(regression) 분석하기¶

2. Prophet 모듈을 이용한 forecast 예측¶

3. Seasonal 시계열 분석으로 주식 데이터 분석하기¶

4. Growth Model¶

1. 관심있는 주식을 분석해 보세요.¶

2. 기간 : 10년치¶

3. 6개월 후 예측해 보세요¶

<한화에어로스페이스 주가예측>¶

'데이터분석' 카테고리의 다른 글

'데이터분석'의 다른글

관련글

티스토리툴바

단축키

내 블로그

블로그 게시글

모든 영역