Indlela Yezikwele Ezincane: Isingeniso kanye Nokusetshenziswa Ekuhlaziyweni Kwedatha
I-Pendahuluan
Indlela Yezikwele Ezincane ingenye yezindlela eziyisisekelo nezisetshenziswa kabanzi ekuhlaziyweni kwedatha, ikakhulukazi kwizibalo kanye nezibalo ezisetshenzisiwe. Le ndlela ihlose ukulinganisa amapharamitha anciphisa isamba sezikwele zokuphambuka okubonwe kumodeli ehlongozwayo. Kulesi sihloko, sizohlola imiqondo eyisisekelo yendlela yezikwele ezincane, ukusetshenziswa kwayo emikhakheni eyahlukene, kanye nezinyathelo ezisebenzayo zokuyisebenzisa.
Imiqondo Eyisisekelo Yendlela Yezikwele Ezincane
Indlela ye-Least Squares ingachazwa kalula ngokusebenzisa i-linear regression. Ake sithi sinedatha ngesimo sama-pairs \((x_i, y_i)\) lapho \( i = 1, 2, …, n \). Imodeli e-linear esifuna ukuyakha ingachazwa kanje:
\[ y = \beta_0 + \beta_1 x + \epsilon \]
lapho \( \beta_0 \) kanye \( \beta_1 \) kuyimingcele esifuna ukuyilinganisela, kuyilapho \( \epsilon \) kuyiphutha noma insalela okulindeleke ukuthi ibe nesilinganiso esingu-zero.
Umgomo wendlela ye-least squares ukunciphisa umsebenzi olandelayo oyinhloko:
\[ S(\beta_0, \beta_1) = \sum_{i=1}^{n} (y_i – \beta_0 – \beta_1 x_i)^2 \]
Ukuthola Amapharamitha Afanele: Indlela Yezibalo
Ukuze sithole amanani epharamitha anciphisa umsebenzi we-objective \( S \), sidinga ukubala ama-derivative angaphelele e-\( S \) maqondana ne-\( \beta_0 \) kanye ne-\( \beta_1 \), bese sixazulula lesi sibalo esilandelayo:
\[ \frac{\partial S}{\partial \beta_0} = -2 \sum_{i=1}^n (y_i – \beta_0 – \beta_1 x_i) = 0 \]
\[ \frac{\partial S}{\partial \beta_1} = -2 \sum_{i=1}^n x_i (y_i – \beta_0 – \beta_1 x_i) = 0 \]
Ngokuxazulula lolu hlelo lwezibalo eziqondile, singathola abaqapheli \(\hat{\beta_0}\) kanye \(\hat{\beta_1}\):
\[ \hat{\beta_1} = \frac{n \sum_{i=1}^n x_i y_i – \sum_{i=1}^n x_i \sum_{i=1}^n y_i}{n \sum_{i=1}^n x_i^2 – (\sum_{i=1}^n x_i)^2} \]
\[ \hat{\beta_0} = \bar{y} – \hat{\beta_1} \bar{x} \]
lapho i-\(\bar{y}\) kanye ne-\(\bar{x}\) kuyizilinganiso ze-\(y\) kanye ne-\(x\) ngokulandelana.
Ukusetshenziswa Kwendlela Yezikwele Ezincane
1. Ezomnotho kanye Nezezimali
Indlela ye-least squares isetshenziswa kabanzi kwezomnotho ukuze ibonise ubudlelwano phakathi kwezinguquko zomnotho. Isibonelo, umhlaziyi wezomnotho angase afune ukukhombisa umthelela wezinga lokungasebenzi ekukhuphukeni kwamanani emali. Esebenzisa indlela ye-least squares, umhlaziyi angakha imodeli yokubuyela emuva ehlobanisa lezi ziguquguquko ezimbili futhi enze iziphetho zezibalo mayelana namandla kanye nohlobo lobudlelwano.
2. Isayensi Yezenhlalo
Kusayensi yezenhlalo, indlela ye-least squares ivame ukusetshenziswa ezinhlolovo kanye nocwaningo lwezengqondo ukutadisha ubudlelwano phakathi kokuziphatha komuntu nezinye izinto eziguquguqukayo. Isibonelo esivamile ukuhlehla okulula okuqondile okuhlobanisa izinga lenjabulo yomuntu nomholo wakhe wonyaka.
3. Ubunjiniyela
Kobunjiniyela, indlela ye-least squares ingasetshenziswa ekulinganisweni kwezinsimbi kanye nokucubungula isignali. Isibonelo, ekucubungulweni kwezithombe zedijithali, le ndlela isetshenziselwa ukunciphisa umsindo ezithombeni ngokufaka imodeli ngokusekelwe kudatha ebonwe.
4. Isayensi Yezulu kanye Nesayensi Yesimo Sezulu
Izazi zezulu zisebenzisa le ndlela ukuhlaziya idatha ehlobene nokushisa, imvula, noma ezinye izinto eziguquguqukayo zesimo sezulu. Ngamamodeli okubuyela emuva, zingabikezela amaphethini esimo sezulu ngokusekelwe kudatha yomlando, okusiza ekudaleni izibikezelo ezinembe kakhudlwana.
Ukusetshenziswa Okusebenzayo ngePython
Ukuze sisebenzise indlela ye-least squares ngokusebenza, ikakhulukazi i-linear regression elula, singasebenzisa ulimi lokuhlela lwe-Python ngosizo lwemitapo yolwazi ethi `numpy` kanye nethi `matplotlib`. Nasi isibonelo sekhodi esibonisa le nqubo:
"`python
ngenisa i-numpy njenge-np
ngenisa i-matplotlib.pyplot njenge-plt
Idatha yesampula
x = np.array([1, 2, 3, 4, 5])
y = np.array([2, 3, 5, 7, 11])
Isilinganiso sika-x no-y
isilinganiso_x = np. isilinganiso(x)
mean_y = np.mean(y)
Bala amapharamitha
inombolo = np.sum((x – mean_x) (y – mean_y))
inani eliyi-denominator = np.sum((x – mean_x) 2)
b1 = inombolo / i-denominator
b0 = mean_y – b1 mean_x
Ukubikezela u-y
y_pred = b0 + b1 x
Imiphumela yesakhiwo
plt.scatter(x, y, umbala='oluhlaza okwesibhakabhaka', ilebula='Idatha Yokubuka')
plt.plot(x, y_pred, color='red', label='Regression Line')
i-plt.xlabel('x')
i-plt.ylabel('y')
i-plt.legend()
i-plt.show()
phrinta(f”Ama-coefficient okubuyela emuva: b0 = {b0}, b1 = {b1}”)
``
Isiphetho
Indlela ye-least squares iyisisekelo esinamandla nesibalulekile ekuhlaziyweni kwezibalo kanye nedatha. Amandla ayo okunciphisa amaphutha kanye nokwandisa ukulingana kwemodeli kuyenza ibe usizo kakhulu emikhakheni ehlukahlukene, kusukela kwezomnotho kuya kobunjiniyela kanye nesayensi yezenhlalo. Nakuba umqondo oyisisekelo ulula, indlela inganwetshwa iye kumamodeli ayinkimbinkimbi kakhulu njenge-nonlinear regression, amamodeli e-mixed-effects, kanye nokufunda komshini. Ngokuqonda kahle indlela ye-least squares kanye nokuzijwayeza okwanele, singathuthukisa ukunemba kokuhlaziywa kwedatha yethu futhi senze izinqumo ezinolwazi oluthe xaxa.
Ngethemba ukuthi lesi sihloko sinikeza umbono ocacile nophelele wendlela ye-least squares kanye nokusetshenziswa kwayo.
