Hanyar Mafi Karancin Murabba'i: Cikakken Nazari
Hanyar Mafi Sauƙi (Least Squares) tana ɗaya daga cikin manyan dabarun nazarin koma-baya na ƙididdiga, tana taka muhimmiyar rawa a cikin daidaita bayanai, ƙira, da nazarin hasashen. Hanya ce ta lissafi da ke rage jimlar murabba'ai na bambance-bambancen da ke tsakanin ƙimar da aka lura da aka annabta. Ta hanyar yin hakan, tana samar da layi ko lanƙwasa mafi dacewa ga wani saitin wuraren bayanai. A cikin wannan labarin, za mu bincika tushe, aikace-aikace, abubuwan da aka samo, da iyakokin Hanyar Mafi Sauƙi (Least Squares).
Historical Background
Carl Friedrich Gauss ne ya fara gabatar da Hanyar Mafi Sauƙi a 1795 yana ɗan shekara 18, kodayake ba a buga ta ba sai a 1809. Adrien-Marie Legendre, wani fitaccen masanin lissafi, ya ƙirƙiro kuma ya buga irin wannan hanyar a 1805. Duk da haka, bayanin Gauss da faɗaɗa hanyar ya kafa harsashin amfani da ita a zamani.
Gidauniyar Lissafi
A cikin mafi sauƙi, ana amfani da Hanyar Mafi Sauƙi don daidaita samfurin layi zuwa saitin wuraren bayanai. Idan aka ba da saitin wuraren bayanai na `n` \((x_1, y_1), (x_2, y_2), …, (x_n, y_n)\, manufar ita ce nemo layi \(y = mx + c\) wanda ke rage jimlar murabba'ai na nisa tsaye (ragowa) tsakanin ƙimar da aka lura \(y_i\) da ƙimar da layin \(\hat{y}_i\) ya annabta.
A fannin lissafi, ana iya wakiltar wannan manufar kamar haka:
\[ S = \sum_{i=1}^{n} (y_i – \hat{y}_i)^2 = \sum_{i=1}^{n} (y_i – (mx_i + c))^2 \]
Wannan lissafi yana wakiltar jimlar murabba'ai na ragowar, \(S\), kuma burinmu shine mu nemo ƙimar \(m\) (gangara) da \(c\) (tsangwama) waɗanda ke rage \(S\).
Kaddamarwa
Domin nemo ƙimar \(m\) da \(c\) waɗanda ke rage jimlar murabba'ai \(S\), mun ɗauki ɓangarorin da suka samo asali na \(S\) dangane da \(m\) da \(c\) muka sanya su zuwa sifili.
1. Tsarin da aka samo daga ɓangaren da ya shafi \(m\):
\[ \frac{\partial S}{\partial m} = \sum_{i=1}^{n} 2(y_i – (mx_i + c))(-x_i) = 0 \]
2. Tsarin da aka samo daga ɓangaren (partial derivative) dangane da \(c\):
\[ \frac{\partial S}{\partial c} = \sum_{i=1}^{n} 2(y_i – (mx_i + c))(-1) = 0 \]
Magance waɗannan daidaiton lokaci guda yana haifar da daidaito na yau da kullun:
\[ m = \frac{n(\sum x_i y_i) – (\sum x_i)(\sum y_i)}{n (\sum x_i^2) – (\sum x_i)^2} \]
\[ c = \frac{(\sum y_i)(\sum x_i^2) – (\sum x_i)(\sum x_i y_i)}{n(\sum x_i^2) – (\sum x_i)^2} \]
Waɗannan daidaito suna ba mu ƙimar \(m\) da \(c\) waɗanda ke rage jimlar ragowar murabba'i.
Aikace-aikace
Hanyar Mafi Sauƙi tana da aikace-aikace iri-iri a fannoni daban-daban:
1. Tattalin Arziki da Kuɗi: Ana amfani da shi don yin ƙira da kuma hasashen alamun tattalin arziki kamar GDP, hauhawar farashin kayayyaki, da farashin hannun jari.
2. Injiniyanci: Ana amfani da shi a sarrafa sigina, tsarin sarrafawa, da gwajin aminci don dacewa da samfura zuwa ga bayanan gwaji.
3. Magani: Yana taimakawa wajen daidaita lanƙwasa girma, lanƙwasa amsawar allurai, da sauran hanyoyin halitta zuwa ga bayanan gwaji.
4. Koyon Inji: Samfuran komawa baya na layi, waɗanda galibi suna amfani da Hanyar Mafi Girman Ma'auni don kimanta sigogi, suna da tushe a cikin dabarun koyo da ake kulawa.
5. Ilimin Taurari: Gauss ya yi amfani da shi a tarihi don ƙididdige kewayar jikin sama.
Bambanci da Tsawo
1. Mafi ƙarancin murabba'ai masu nauyi: A lokuta inda lura ke da bambance-bambance daban-daban, Mafi ƙarancin murabba'ai masu nauyi suna sanya nauyi ga kowane ma'aunin bayanai, suna rage jimlar ma'aunin murabba'ai.
2. Ƙananan murabba'ai marasa layi: Yana faɗaɗa hanyar don dacewa da samfuran da ba na layi ba, sau da yawa ta hanyar dabarun maimaitawa kamar algorithms na Gauss-Newton ko Levenberg-Marquardt.
3. Mafi ƙarancin murabba'ai (Rage Rage Ridge): Yana ƙara wa jimlar murabba'ai hukunci don hana wuce gona da iri, musamman ma idan ana maganar multicollinearity.
4. Manyan Ƙananan Murabba'ai: Yana lissafin abubuwan da aka lura da su, yana faɗaɗa hanyar ƙananan murabba'ai don magance kurakuran da suka shafi kai tsaye da kuma waɗanda ba su da alaƙa da juna.
Kayan Aiki
Kayan aikin software daban-daban da yanayin shirye-shirye suna ba da ayyuka na ciki don yin mafi ƙarancin juzu'i na murabba'i:
1. Python: Laburare kamar NumPy da SciPy suna ba da ayyuka `numpy.linalg.lstsq` da `scipy.optimize.curve_fit` don daidaitawar mafi ƙarancin murabba'ai masu layi da marasa layi, bi da bi.
2. R: Ayyuka kamar `lm()` don samfuran layi da `nls()` don samfuran marasa layi a cikin cikakken yanayin ƙididdiga na R.
3. MATLAB: Ana amfani da aikin `lsqcurvefit` a cikin MATLAB don daidaitawa mara layi, tare da tallafi a ciki don ayyukan algebra na layi.
gazawar
Duk da fa'idar amfaninta, Hanyar Mafi Sauƙi tana da iyakoki:
1. Jin Daɗi ga Masu Fita: Ƙananan Squares suna da matuƙar jin daɗi ga masu fita, saboda kalmar murabba'i tana ƙara tasirin manyan ragowar.
2. Zaton Layi: Yana ɗaukar alaƙar layi tsakanin masu canji, wanda ƙila ba zai yi daidai da bayanai masu rikitarwa ba.
3. Matsalolin Multicollinearity: A cikin koma-baya na layi da yawa, multicollinearity tsakanin masu hasashen abubuwa na iya karkatar da kimantawar ma'auni.
4. Daidaito tsakanin Bambance-bambance (Homoscedasticity): Yana ɗauka cewa bambancin kalmomin kuskure yana nan a kowane lokaci a cikin lura, wanda ƙila ba koyaushe haka lamarin yake ba.
Kammalawa
Hanyar Mafi Girman Ma'auni, duk da shekarunta, ta kasance ginshiƙi a cikin nazarin bayanai da ƙira. Sauƙinta, ingancinta, da daidaitawa ga siffofi daban-daban, tun daga koma-baya na layi zuwa samfura masu rikitarwa marasa layi da na gabaɗaya, sun sa ta zama mai mahimmanci. Fahimtar tushen lissafi, aikace-aikacenta, da iyakokinta yana da mahimmanci ga duk wanda ke da hannu a cikin nazarin bayanai, tabbatar da cewa an yi amfani da hanyar yadda ya kamata don samun fahimta mai ma'ana. Ko a cikin ilimi ko masana'antu, Hanyar Mafi Girman Ma'auni ta ci gaba da zama kayan aiki mai ƙarfi don fahimtar alaƙar da ke cikin bayanai da kuma yin hasashen da ya dace.
