Isikhathi Series and Trend ukuhlaziya inselelo ephefumulelwe ama-datasets omhlaba wangempela

Isithombe ngombhali | Ikantheyili

Okusobala Ukuqalisa

Idatha yochungechunge yesikhathi ikuyo yonke indawo. Amanani esitoko agxuma nsuku zonke. Amazinga okushisa afuduka. Iwebhusayithi Spikes ye-WEBSITE NOKUQHAWULA. Iningi labantu lihlela umugqa. Bese bema.

Kepha nakhu ishadi elilodwa ngeke likutshele: Ngabe ukuthambekela kuyashesha? Yehlisa ijubane? Cishe ukubuyisa ngokuphelele?

Kulesi sihloko, sizohlaziya okulindelwe ukwehla kwamandla emali esebenzisa amasu amathathu ahambisanayo: Ukuhambisa Izilinganiso, Izinguquko Zonyaka Wonyakafuthi Amabhendi we-bollinger.

Isithombe nguMlobi

Indlela ngayinye iphendula umbuzo ohlukile mayelana nedatha efanayo. Izilinganiso ezihambisayo zembula indlela yokuqondisa, ukuguquguquka konyaka onyaka kakhulu ukugqamisa amashifu asemanzini, kanye namabhande oku-bollinger aveza izikhathi zokunyakaza okwedlulele.

Sizosebenzisa lawo masu ukuhlaziya umkhuba weminyaka emi-5 yeminyaka emi-5, kusuka ngo-Okthoba 2020 kuya ku-Okthoba 2025.

Okusobala Ukuqonda idatha yethu: ukudicilela phansi isilinganiso seminyaka eyi-10 yokungenisa amandla

Ukuqonda idatha yethu, okokuqala kudingeka siqonde i-metric eyakhelwe ku: Inani leminyaka eyishumi le-freaseven (T10yie).

I-T10yie imele okulindelwe ukwehla kwamandla eMakethe eminyakeni eyishumi ezayo. I-Math elula: Ukhipha i-inflation-evikelekile ukuvezwa kwezithelo zemali ezivikelekile ezivuneni zoMgcinimafa ojwayelekile.

// Kusho ukuthini?

Uma i-t10yie = 2,5%, imakethe ilindele ukwehla kwamanani okuphakathi kwe-2,5% ngaphezulu kweminyaka eyi-10. Amanani aphezulu asho okulindelwe ukwehla kwamandla emali. Amanani aphansi asho ukwehla kwamandla emali okubuthakathaka noma ukwesaba kwe-deflation.

// Kungani ososayensi bezomnotho kanye ne-Fed Bukela Leli Cares Like Like Haleks

I-Federal Reserve ibuka le metric eduze. I-Rising Breakeven ilele ukukhathazeka kwamanani entengo okubonakalayo okungase kubangele Izinga lenzalo ye-Federal Reserve's Hikes. Amaconsi abukhali angakhombisa ukwesaba komnotho noma izingcindezi zokuhlehlisa.

// Idatha yethu shazi: Iminyaka engu-5 yokulindela ukwehla kwamandla emali (2020-2025)

Manje sizosebenzisa leli dataset.

Iskrini | Itred

Chofoza ku- “Landa” ukuze ugcine ifayela emshinini wakho.

Uma unesifiso sokuhlola imininingwane efanayo yangempela yomhlaba kanye nokuhlaziywa kwedatha nokubuka ngeso, hlola Stratascratch. Kuyipulatifomu yokuthola ama-datasets angamaqiniso asetshenziswa kuzo zonke izimali, ubuchwepheshe kanye nemithombo yedatha yomphakathi.

// Ukujwayela imininingwane: Ukwakheka, umthombo, kanye nezibalo ezifingqiwe

Nalu ulwazi oluthile mayelana dataset yethu:

• Umthombo: Idatha yezomnotho ye-Federal Reserve (FRED).
• Isikhathi: Okthoba 2020 – Okthoba 2025 (iminyaka emi-5).
• Imvamisa: Ukubonwa kwansuku zonke.
• Ukubonwa okuphelele: Amaphoyinti wedatha ayi-1,305.
• Ibanga: 1.64% kuya ku-3.02%.
• Isilinganiso: 2.33%.

Masifunde le datha bese ubona imigqa embalwa yokuqala. Nayi ikhodi:

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

df= pd.read_csv("T10YIE.csv")
df.head()

Nansi umphumela:

Kuyindlela elula yedatha, equkethe amakholomu amabili kuphela: observation_date na- T10YIE.

Okusobala Ukuhlaziywa kweThendi: amasu amathathu wesikhathi sochungechunge lwesikhathi

Sizoqala ngenqubo yokuhamba eshukumisayo.

// I-Techyique 1: Izilinganiso Zokuhamba

Ukuhambisa izilinganiso ezisheshayo ukuguquguquka kwesikhashana kwesikhashana. Baveza ukuthambekela okungaphansi. Thatha isilinganiso seminyaka engama-30. Ibala ukuthi kusho ukuthini izinsuku ezingama-30 ezedlule. Umphumela? Umugqa onobuhlakani obuhlunga umsindo wansuku zonke.

Izimakethe zezimali ziyisiphithiphithi. Amanani ansuku zonke ahamba ngezihloko zezindaba. Bayeka imibiko etholwayo. Imicimbi ye-Geopolitical ibathumela emaceleni. Izilinganiso ezihambayo zisikwe kukho konke lokhu. Bakukhombisa indlela yangempela yokuqondisa ngaphansi kwezinxushunxushu.

Izinhlobo:

• I-STORE-term ma (izinsuku ezingama-30): Ibamba amashifu asanda kwenzeka.
• I-MA-ter-term ma (izinsuku ezingama-90): ikhombisa indlela ebanzi yokuqondisa.
  • Ama-Crossovers: lapho ama-marm amafushane awela ngaphezulu kwe-MA ende.

Nayi ikhodi:

df['T10YIE'] = df['T10YIE'].ffill()
df['MA_30'] = df['T10YIE'].rolling(window=30).mean()
df['MA_90'] = df['T10YIE'].rolling(window=90).mean()

plt.figure(figsize=(15, 7))

plt.plot(df.index, df['T10YIE'], label="Daily Rate", alpha=0.4, linewidth=0.8, color="gray")
plt.plot(df.index, df['MA_30'], label="30-Day MA", linewidth=2, color="blue")
plt.plot(df.index, df['MA_90'], label="90-Day MA", linewidth=2, color="red")

plt.axvspan(0, 200, color="palegreen", alpha=0.3, label="Phase 1: Recovery")
plt.axvspan(200, 500, color="lightcoral", alpha=0.3, label="Phase 2: Volatility")
plt.axvspan(500, 1000, color="lightblue", alpha=0.3, label="Phase 3: Decline")
plt.axvspan(1000, df.index[-1], color="plum", alpha=0.3, label="Phase 4: Stabilization")

plt.title('Breakeven Inflation Rate with Highlighted Phases', fontsize=14, fontweight="bold")
plt.ylabel('Inflation Rate (%)')
plt.xlabel('Date')
plt.grid(True, alpha=0.3)

plt.legend(loc="upper right")

plt.tight_layout()
plt.show()

Nansi umphumela:

// Imiphumela & Ukuhunyushwa

Izilinganiso ezishukumisayo ziveza amaphethini ahlukile kuwo wonke iminyaka emihlanu yokulindela ukwehla kwamandla emali.

Isigaba 1: Ukubuyiselwa okubukhali (izinsuku 0-200)
Zombili izilinganiso zikhuphuka ngokuqinile kusuka ku-1.7% kuya ku-2.4%. I-Maiden mari ikhuphuka ngokushesha. Lesi sikhathi sidonsa i- Ukuvulwa kabusha kwe-Post-Covid Economid. -Yisimuhluza ukukhuthaza imali Ukutholwa okulindelwe kwe-inflation kuya phezulu.

Isigaba 2: Isikhathi esiphakeme sokuqina (izinsuku 200-500)
Amanani ansuku zonke afinyelela ku-3.0% ezungeze usuku lwe-400. I-30-day mai ifinyelela ku-2.9%. Lokhu kufana 2022 Ukuhlinzwa Kwenflation. Ukuphazamiseka Kwe-Supply Chain hit. I-Russia ihlasele i-Ukraine. Kuqhume amanani entengo yamandla.

Isigaba 3: Ukwehla (Izinsuku 500-1000)
Izinsuku ezingama-30 zezinsuku ezingama-30 zeza phansi kakhulu, zehlela ku-2.2% eduze nosuku lwe-1000. Amanani entengo ahamba ngezinyawo anolaka kuwo wonke ama-2022 naku-2023. Ukulindelwa kwe-inflation kupholile njengoba inqubomgomo isebenza.

Isigaba 4: Ukuqina kwakamuva (izinsuku 1000-1300)
Ama-mailou ama-mali angama-30 azungeze ama-2.3% afinyelela ku-2.4%. Ukuguquguquka okuncane. Izimakethe zibonisa ukuzethemba ukuthi ukwehla kwamandla emali kuvame ukubekezelela eduze kwe Ithagethi engu-2% ye-Fed. Ukukala izintaba zimise okwesikhashana.

Ukuqonda Okusemqoka
I-Maifa lezinsuku ezingama-30 yathola yonke indawo yokuguqula. Lapho ekhuphuka kakhulu ekuqaleni kwawo-2021, kwalandelwa ukwanda kwe-inflation. Lapho yehla maphakathi no-2022, kwaqala ukupholisa. Ukuqina kwamanje kuphakamisa izimakethe ukholelwa ukuthi ukwethuka kwamanani sekudlule.

// I-Technique 2: Ushintsho olwedlula unyaka nonyaka

Ushintsho lonyaka olwedlule (i-YOY) luqhathanisa inani lanamuhla ngosuku olufanayo ngonyaka owodwa edlule. Iyaphendula: “Ingabe ukulindelwa kwamanani okude kuphakeme noma kuphansi kunale bekuyizinyanga eziyi-12 ezedlule?”

Lokhu kususa umsindo wezinkathi ezithile futhi kukhombisa umfutho oqondile. Amanani amahle = okulindelwe unyaka okhuphuka unyaka. Amanani amabi = okulindelwe ukuwela unyaka nonyaka. Zero = inkambiso eyisicaba.

Nayi ifomula yokubala i-Yoy Change, lapho (v_t ) inani lamanje futhi (v_ {{t-365} ) inani elivela ngonyaka owodwa (cishe izinsuku ezingama-252 zokuhweba) edlule:

$ $
Umbhalo {yoy change} = v_t – v_ {t-365}
$ $

Kwikhodi, kubukeka kanjena:

import pandas as pd
import matplotlib.pyplot as plt

df['T10YIE'] = df['T10YIE'].ffill()
# Calculating diff based on trading days (approx 252 per year)
df['YoY_Change'] = df['T10YIE'].diff(252)

fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(14, 10), sharex=True)

ax1.plot(df.index, df['T10YIE'], color="blue", linewidth=1)
ax1.set_ylabel('Inflation Rate (%)')
ax1.set_title('Breakeven Inflation Rate (Original)', fontsize=12, fontweight="bold")
ax1.grid(True, alpha=0.3)

ax2.plot(df.index, df['YoY_Change'], color="darkred", linewidth=1.5)
ax2.axhline(y=0, color="black", linestyle="--", linewidth=1.5, alpha=0.7)

ax2.fill_between(df.index, df['YoY_Change'], 0,
    where=(df['YoY_Change'] > 0), color="green", alpha=0.3, label="Rising YoY")
ax2.fill_between(df.index, df['YoY_Change'], 0,
    where=(df['YoY_Change'] <= 0), color="red", alpha=0.3, label="Falling YoY")
ax2.set_ylabel('YoY Change (%)')
ax2.set_xlabel('Date')
ax2.set_title('Year-over-Year Change in Inflation Expectations', fontsize=12, fontweight="bold")
ax2.grid(True, alpha=0.3)

# First Green Zone (Days 250-500)
ax1.axvspan(250, 500, color="palegreen", alpha=0.4, label="First Green Zone")
ax2.axvspan(250, 500, color="palegreen", alpha=0.4)

# Red Zone (Days 500-1000)
ax1.axvspan(500, 1000, color="lightcoral", alpha=0.4, label="Red Zone")
ax2.axvspan(500, 1000, color="lightcoral", alpha=0.4)

# Second Green Zone (Days 1000-1300)
ax1.axvspan(1000, df.index[-1], color="mediumaquamarine", alpha=0.4, label="Second Green Zone")
ax2.axvspan(1000, df.index[-1], color="mediumaquamarine", alpha=0.4)

ax1.legend(loc="upper left")
ax2.legend(loc="upper left")

plt.tight_layout()
plt.show()

Nansi umphumela:

// Imiphumela & Ukuhunyushwa

Ishadi lokushintsha le-Yoy lihlukanisa okulindelwe ama-inflation ibe yizindawo eziluhlaza nezibomvu. Kusho ukushesha. Okubomvu kusho ukudedela. Lokhu kwembula ukuthi umfutho ushintsha ishadi lesilinganiso sokuqala esiphuthelwe ngokuphelele.

Indawo yokuqala eluhlaza (izinsuku 250-500)
Ukulindelwa kwe-inflation kukhuphuke ngokushesha. Izinguquko zonyaka owedlule ziye zangena ku- + 1.0%. Le nkathi? 2021 kuya ku-2022. Ukuhambisa amaketanga kudilikile. Amasheke okuvuselela akhukhula umnotho. I-Russia ihlasele i-Ukraine. Kuqhume amanani entengo yamandla.

I-Red Zone (Izinsuku 500-1000)
Okulindelwe kwaphahlazeka. Bawela ku -0.75% unyaka owodwa. Amanani e-Federal Reserve Hiked ngenhlaka kube ngu-2022 no-2023. Izimakethe ezikholelwa ukuthi ukwehla kwamandla emali kungaphola. Babelungile.

Indawo yesibili eluhlaza (izinsuku 1000-1300)
Izinguquko ezincane ezibunjiwe zibuyile. Bahlaselwa phakathi kuka-0,1% no-0.3%. Okulindelwe kuyeka ukuwa. Baqale ukuzilimaza ngenhla konyaka owedlule. Lokhu kusayina okujwayelekile, hhayi ukwethuka.

Isiginali ngekusasa
Iziqeshana zakamuva eziluhlaza zimnene ziqhathaniswa nokuhlinzwa okungu-2022. Izinguquko ze-Yoy ezingezansi + 0.25%? Okulindelwe kuhlala kugxiliwe. Ukunyakaza okuqhubekayo ngenhla + 0.5%? Lokho bekuzohlabelela ukukhathazeka kwamanani okukhathaza okukufanele ukubukwa.

// I-Techturique 3: Ama-Bollinger Band (imvilophu yokushintshanisa)

Ama-Bollinger band adala umngcele ongenhla naphansi ozungeze isilinganiso esinyakazayo usebenzisa ukuphambuka okujwayelekile. Amabhendi ayanda ngezikhathi eziguqukayo kanye nenkontileka ngesikhathi sokuthula.

Kubonisa lapho okulindelwe ukwehla kwamanani “okujwayelekile” (amabhendi angaphakathi) nokuqhathanisa “okwedlulele” (amabhendi angaphandle). Lapho izinga lithinta ibhendi ephezulu, liphakeme ngokungajwayelekile. Lapho ithinta ibhendi ephansi, iphansi ngendlela engajwayelekile.

Isakhiwo:

• I-Middle Band: Isilinganiso esinyakazayo sezinsuku ezingama-20.
• I-Band ephezulu: Middle + (2 × ukuphambuka okujwayelekile).
• I-Band Band: Middle – (2 × ukuphambuka okujwayelekile).

Ibanga elidaliwe lisho ukuthi ama-95% wedatha kufanele awele ngaphakathi kwamabhande. Lokhu kungavezwa ngokusemthethweni njengoba:

$ $
Umbhalo {Upper} = mu_ {20} + (2 times sigma_ {20})
$ $
$ $
Umbhalo {ngaphansi} = mu_ {20} – (2 times sigma_ {20})
$ $

Nayi ikhodi:

df['T10YIE'] = df['T10YIE'].ffill()

window = 20
df['BB_Middle'] = df['T10YIE'].rolling(window=window).mean()
df['BB_Std'] = df['T10YIE'].rolling(window=window).std()
df['BB_Upper'] = df['BB_Middle'] + (2 * df['BB_Std'])
df['BB_Lower'] = df['BB_Middle'] - (2 * df['BB_Std'])

plt.figure(figsize=(15, 7))

plt.plot(df.index, df['T10YIE'], label="Daily Rate", color="black", linewidth=0.8)
plt.plot(df.index, df['BB_Middle'], label="20-Day MA", color="blue", linewidth=1.5)
plt.plot(df.index, df['BB_Upper'], label="Upper Band", color="red", linewidth=1, linestyle="--")
plt.plot(df.index, df['BB_Lower'], label="Lower Band", color="green", linewidth=1, linestyle="--")
plt.fill_between(df.index, df['BB_Upper'], df['BB_Lower'], alpha=0.1, color="gray")

plt.axvspan(350, 450, color="gold", alpha=0.3, label="Band Expansion (Volatility↑)")
plt.axvspan(800, 1200, color="lightblue", alpha=0.3, label="Band Contraction (Volatility↓)")

plt.axvspan(190, 210, color="lightcoral", alpha=0.5, label="Upper Breach (~Day 200)")
plt.axvspan(390, 410, color="lightcoral", alpha=0.5, label="Upper Breach (~Day 400)")

plt.axvspan(1040, 1060, color="palegreen", alpha=0.5, label="Lower Touch (~Day 1050)")

plt.title('Breakeven Inflation Rate with Bollinger Bands & Key Events', fontsize=14, fontweight="bold")
plt.ylabel('Inflation Rate (%)')
plt.xlabel('Date')
plt.grid(True, alpha=0.3)
plt.legend(loc="upper left")

plt.tight_layout()
plt.show()

Nansi umphumela:

// Imiphumela & Ukuhunyushwa

Amabhendi we-bollinger akhomba lapho okulindelwe ukwehla kwamandla emali ekweqisayo okujwayelekile.

Ukunwetshwa kwe-Band (Izinsuku 350-450)
Amabhendi anwebekile ngokumangazayo njengoba izinga le-Daily Live liphindaphinda ibhendi engenhla, eshaya ama-3.0%. Lesi sikhathi sathumba 2022 ukwethuka inflation Ngesikhathi I-Russian-Ukraine War lapho ukuguquguquka kwemakethe kudlula.

Ukwephulwa kwe-band ephezulu
Ukuthinta okuningi kwebhendi ephezulu (izinsuku ezingama-200, 400) ukwethuka kwezimakethe zesiginali, okulindelwe kugxume ngaphezu kwamabanga ajwayelekile. Ukwephulwa ngakunye kuxwayise ngokuthi ukwesaba kwamanani entengo kuyashesha.

I-Band Contraction (Izinsuku 800-1200)
Amabhendi anciphisa kakhulu ngezinga elihlala ngaphakathi. Lokhu kukhombisa ukuguquguquka kuwele njengoba Izinga lokulinganisa leFed lisebenze futhi izimakethe zifinyelelwe kuvunyelwe.

I-Band Band touch (Usuku 1050)
Isilinganiso sishaye ibhendi ephansi ku-2.05%, ukusayina ngethemba elingajwayelekile ngesikhathi sekwephuzile-2023 Ukwesaba ukwehla kabusha.

Isiginali ngekusasa
Amabhande amancane anciphile kanye nezinga elizinzile (2.35%) akhombisa indlela evamile yokuziphatha kwemakethe. Ukwephulwa okusha kwe-band ephezulu ngenhla kwe-2,5% kuzobe kuvuselela ukukhathazeka kwamanani entengo.

Okusobala Amasu ahlukene, izindaba ezihlukile

Ukuhlaziywa kweThendi akuphathelene nokubikezela ikusasa; Kumayelana nokuqonda ukuthi idatha ikutshelani. Inani lamanani entengo yeminyaka eyi-10 ukusuka ku-2020 kuya ku-2025 kwembula amaphethini ahlukile usebenzisa inqubo ngayinye.

Noma imicimbi yomhlaba wonke efana nokuhlasela kweRussia-Ukraine noma inkinga yamandla ithinta konke ukuhlaziya, inqubo ngayinye ichaza umthelela wawo ngendlela ehlukile. Isilinganiso esishukumisayo singakhombisa ukuhamba kancane kancane, ushintsho lonyaka olwedlule lungaqhakambisa ukuswayipha okubukhali, ngenkathi amabhendi we-bollinger angahle ahlele isikhathi esifanayo nesipikili esiguquguqukayo.

Kungakho ukukhetha izindlela zakho zokuhlaziya izindlela zokuhlaziya; Ibunjwa ukuthi uyibona kanjani le ndaba kwidatha yakho. Umcimbi ofanayo ungabukeka njengokululama, ukungazinzi, noma okujwayelekile ngokuya ngelensi yokuhlaziya oyisebenzisayo.

Okusobala Ukugcina

Isifundo sangempela akuyona yiphi inqubo enhle kakhulu; Ukwazi ukuthi uzosebenzisa nini. Lezi zindlela ezintathu zisebenza ngamanani esitoko, ithrafikhi yewebhu, idatha yokuthengisa, noma yini ehamba ngokuhamba kwesikhathi. Amaphethini akhona. Udinga nje amathuluzi afanele ukuze ubabone.

Ngamanye amagama, imininingwane ayivami ukukhuluma ngezwi elilodwa. Indlela oyikhethayo inquma umyalezo owuzwayo. Kungakho ukuhlaziya kwethayela kungokuchazwa njengoba kumayelana nokubalwa.

Nate rosidi Ungusosayensi wedatha kanye necebo lomkhiqizo. Ungumuntu futhi ongumsunguli wokufundisa ofundisayo, futhi umsunguli weStratascratch, isiteji esisiza ososayensi bedatha balungiselele izingxoxo zabo ngemibuzo yangempela yengxoxo evela ezinkampanini eziphezulu. UNate ubhala ezitayeleni zakamuva emakethe yezemisebenzi, unikeza izeluleko ezixoxiswayo, wabelana ngamaphrojekthi weSayensi yedatha, futhi amboze yonke into SQL.