Zviri Mukati[Viga][Ratidza]
Kana iwe uri Python programmer kana iwe uchitsvaga ine simba toolkit yekushandisa kusuma muchina kudzidza mukugadzira system, Scikit-dzidza iraibhurari yaunofanirwa kutarisa.
Scikit-Lear yakanyorwa zvakanaka uye iri nyore kushandisa, ungave uri mutsva pakudzidza muchina, unoda kusimuka nekumhanya nekukasira, kana kuda kushandisa chishandiso cheML chekutsvagisa.
Inokutendera iwe kuti uvake yekufungidzira data modhi mumitsetse mishoma yekodhi uye wozoshandisa iyo modhi kuti ienderane nedata rako seraibhurari yepamusoro. Inochinjika uye inoshanda zvakanaka nemamwe Python raibhurari seMatplotlib yekuchati, NumPy yearray vectorization, uye pandas yekuona data.
Mugwaro iri, iwe unowana zvese nezve kuti chii, mashandisiro aungaite, pamwe nezvayakanakira nezvayakaipira.
Chii Scikit-dzidza?
Scikit-Learn (inozivikanwawo se sklearn) inopa dzakasiyana siyana seti yemamodhi uye kudzidza muchina. Kusiyana nemamodule akawanda, sklearn inogadzirwa muPython panzvimbo yeC. Pasinei nekugadzirwa muPython, kushanda kwe sklearn kunoratidzwa nekushandiswa kwayo kweNumPy nokuda kwepamusoro-performance linear algebra uye array operations.
Scikit-Learn yakagadzirwa sechikamu cheGoogle's Zhizha reCode purojekiti uye kubvira zvaita kuti hupenyu hwemamiriyoni ePython-centric data masayendisiti pasirese ave nyore. Ichi chikamu cheiyi nhevedzano chakanangana nekupa raibhurari uye kutarisa pane chimwe chinhu - dhataset shanduko, inova yakakosha uye yakakosha nhanho yekutora usati wagadzira yekufungidzira modhi.
Raibhurari yakavakirwa paSciPy (Scientific Python), iyo inofanirwa kuiswa usati washandisa scikit-dzidza. Iyi stack ine zvinhu zvinotevera:
- NumPy: Python's standard n-dimensional array package
- SciPy: Iyo yakakosha pasuru yesainzi komputa
- Pandas: Data zvimiro uye kuongorora
- Matplotlib: Iyo ine simba 2D/3D yekuronga raibhurari
- Sympy: Symbolic masvomhu
- IPython: Yakavandudzwa inopindirana console
Zvishandiso zveraibhurari yeScikit-dzidza
Scikit-dzidza ndeye yakavhurika-sosi Python package ine sophisticated data data uye migodhi maficha. Iyo inouya nehuwandu hweakavakirwa-mukati maalgorithms ekukubatsira iwe kuwana zvakanyanya kubva kune yako data sainzi mapurojekiti. Raibhurari yeScikit-inodzidza inoshandiswa nenzira dzinotevera.
1. Kudzvinyirira
Regression analysis inyanzvi yenhamba yekuongorora nekunzwisisa kubatana pakati pezviviri kana kupfuura. Nzira inoshandiswa kuita regression ongororo inobatsira pakuona kuti ndezvipi zvinhu zvine basa, izvo zvinogona kufuratirwa, uye kuti zvinodyidzana sei. Maitiro ekudzoreredza, semuenzaniso, anogona kushandiswa kunzwisisa zviri nani maitiro emitengo yemasheya.
Regression algorithms inosanganisira:
- Kudzokorora Kwemashoko
- Ridge Regression
- Kudzora kweLasso
- Chisarudzo Muti Regression
- Random Sango
- Tsigira Vector Machines (SVM)
2. Kupatsanura
Nzira yeKupatsanura inzira Yakatariswa Yekudzidza iyo inoshandisa data rekudzidzisa kuona chikamu chezvekucherekedza patsva. An algorithm muClassification inodzidza kubva kune yakapihwa dhatabheti kana zvakacherechedzwa uye wozoisa kumwe kucherechedzwa mune imwe yemakirasi akawanda kana mapoka. Ivo vanogona, semuenzaniso, kushandiswa kurongedza email kutaurirana se spam kana kwete.
Classification algorithms inosanganisira zvinotevera:
- Kudzvinyirira Kunogadzirisa
- K-Vavakidzani Vepedyo
- Tsigira Vector Machine
- Sarudzo Muti
- Random Sango
3. Kubatanidza
Iwo clustering algorithms muScikit-dzidza anoshandiswa kuronga otomatiki data rine zvivakwa zvakafanana mumaseti. Kukokorodza (clustering) inzira yekuisa zvinhu muboka kuitira kuti avo vari muboka rimwechete vafanane nezviri mune mamwe mapoka. Mutengi data, semuenzaniso, inogona kupatsanurwa zvichienderana nenzvimbo yavo.
Clustering algorithms inosanganisira zvinotevera:
- DB-SCAN
- K-Means
- Mini-Batch K-Zvinoreva
- Spectral Clustering
4. Muenzaniso Kusarudzwa
Kusarudza maalgorithms emuenzaniso anopa nzira dzekuenzanisa, kusimbisa, uye kusarudza iwo akakwana maparamita uye modhi dzekushandisa mumatanho esainzi yedata. Yakapihwa data, kusarudzwa kwemuenzaniso ndiro dambudziko rekutora nhamba yemuenzaniso kubva kuboka revanoda kukwikwidza. Mumamiriro ezvinhu akanyanya, kuunganidzwa kwe data kusati kwavapo kunotariswa. Zvisinei, basa racho rinogonawo kusanganisira kugadzirwa kwezviedzo kuitira kuti data yakawanikwa inonyatsokodzera dambudziko rekusarudza muenzaniso.
Mamodule ekusarudza modhi anogona kuvandudza kurongeka nekugadzirisa paramita anosanganisira:
- Muchinjikwa-kusimbisa
- Grid Search
- Metrics
5. Dimensionality Reduction
Kuendeswa kwedata kubva panzvimbo yepamusoro-soro kuenda kune yakaderera-dimensional nzvimbo kuitira kuti iyo yakaderera-dimensional inomiririra kuchengetedza zvimwe zvakakosha zve data rekutanga, zvine hungwaru padyo nechiyero chayo chekuzvarwa, inozivikanwa sekuderedzwa kwedimensionality. Nhamba yezvinyorwa zvisinganzwisisiki zvekuongorora zvinoderedzwa kana chiyero chakaderera. Data rekunze, semuenzaniso, rinogona kunge risingatariswe sekuvandudza kugona kwekuona.
Dimensionality Reduction algorithm inosanganisira zvinotevera:
- Feature sarudzo
- Chikuru Chikamu Chekuongorora (PCA)
Kuisa Scikit-dzidza
NumPy, SciPy, Matplotlib, IPython, Sympy, uye Pandas inodiwa kuti iiswe usati washandisa Scikit-dzidza. Ngatiiisei tichishandisa pip kubva kuconsole (inoshanda chete kuWindows).
Ngatiise Scikit-dzidza izvozvi kuti takaisa maraibhurari anodiwa.
Features
Scikit-dzidza, dzimwe nguva inozivikanwa se sklearn, iPython toolkit yekushandisa michina yekudzidza modhi uye nhamba yekuenzanisa. Isu tinogona kuishandisa kugadzira akawanda emuchina ekudzidza modhi yekudzoreredza, kupatsanura, uye kubatanidza, pamwe nematurusi ezviverengero ekuongorora aya mamodheru. Inosanganisirawo kuderedzwa kwedimensionality, kusarudzwa kwechimiro, kuburitsa maficha, ensemble nzira, uye akavakirwa-mukati dhatasethi. Tichaongorora chimwe nechimwe cheunhu uhwu chimwe panguva.
1. Kupinza Datasets
Scikit-Learn inosanganisira akati wandei ekare-akavakwa dhataseti, senge iris dataset, mutengo wepamba dhata, titanic dataset, zvichingodaro. Mabhenefiti akakosha emaseti aya ndeekuti ari nyore kubata uye anogona kushandiswa kukurumidza kugadzira mhando dzeML. Aya ma dataset akakodzera kune vatsva. Saizvozvo, unogona kushandisa sklearn kuunza mamwe ma dataset. Saizvozvo, unogona kuishandisa kuunza mamwe ma dataset.
2. Kuparadzanisa Dataset yeKudzidzisa uye Kuedza
Sklearn yaisanganisira kugona kugovera iyo dataset kuita kudzidziswa uye kuyedza zvikamu. Kupatsanura dhatabheti kunodiwa pakuongorora kwakasarudzika kwekufanotaura kuita. Isu tinogona kutsanangura kuti yakawanda sei data yedu inofanirwa kuverengerwa muchitima uye bvunzo dataset. Isu takapatsanura dataset tichishandisa chitima test split zvekuti chitima seti chinosanganisira 80% yedata uye test set ine 20%. Iyo dataset inogona kukamurwa sezvinotevera:
3. Linear Regression
Linear Regression inotarisirwa kudzidza-yakavakirwa muchina kudzidza maitiro. Inoita basa rekudzoreredza. Kubva pane zvakazvimiririra zvakasiyana, regression modhi ine chinangwa chekufanotaura kukosha. Inonyanya kushandiswa kuona kubatana pakati pezvinosiyana uye kufanotaura. Mhando dzakasiyana dzekudzosa dzinosiyana maererano nemhando yekubatanidza yavanoongorora pakati pezvinotsamira uye zvakazvimiririra, pamwe nenhamba yezvakazvimiririra zvinosiyana zvinoshandiswa. Isu tinogona kungogadzira iyo Linear Regression modhi tichishandisa sklearn sezvinotevera:
4. Logistic Regression
A common categorization approach is logistic regression. Iri mumhuri imwechete sepolynomial uye mutsara regression uye ndeye mutsara classifier mhuri. Zvakawanikwa zveLogistic regression zviri nyore kunzwisisa uye zvinokurumidza kuverenga. Nenzira imwechete sekudzoreredza kwemutsara, logistic regression inzira inotariswa yekudzoreredza. Iyo inobuda yakasiyana ndeye categorical, saka ndiyo chete mutsauko. Inogona kuona kana murwere ane chirwere chemwoyo kana kwete.
Nyaya dzakasiyana-siyana dzekuisa, dzakadai sekuona spam, dzinogona kugadziriswa uchishandisa logistic regression. Kufembera kweshuga, kuona kana mutengi achatenga chimwe chigadzirwa kana chinja kune mukwikwidzi, achitarisa kana mushandisi achadzvanya pane chaiyo yekushambadzira link, uye zvimwe zvakawanda zvimwe zvingori mienzaniso mishoma.
5. Muti Wechisarudzo
Iyo yakanyanya simba uye yakashandiswa zvakanyanya kurongedza uye kufanotaura maitiro ndiwo muti wesarudzo. Muti wesarudzo chimiro chemuti chinoratidzika sechati chinoyerera, chine nodi yega yega yemukati inomiririra bvunzo pane hunhu, bazi rega rega rinomiririra mhedziso yebvunzo, uye imwe neimwe yeshizha node (terminal node) yakabata kirasi label.
Kana izvo zvinotsamira zvisina hukama hwemutsara neakazvimiririra akasiyana, kureva kana mutsara regression isingaburitse zvakawanikwa, miti yesarudzo inobatsira. Iyo DecisionTreeRegression () chinhu chinogona kushandiswa nenzira yakafanana kushandisa muti wesarudzo kuti udzoke.
6. Random Sango
Sango risina kurongeka ndiro a machine learning nzira yekugadzirisa kudzoreredza uye kurongedza nyaya. Inoshandisa ensemble kudzidza, inova nzira inosanganisa akawanda ekirasi kugadzirisa matambudziko akaomarara. Nzira yesango isina kurongeka inoumbwa nenhamba yakawanda yemiti yesarudzo. Inogona kushandiswa kurongedza zvikumbiro zvechikwereti, kuona hunyengeri, uye kutarisira kubuda kwechirwere.
7. Kuvhiringidzika Matrix
A confusion matrix tafura inoshandiswa kutsanangura maitiro emhando yemhando. Aya mazwi mana anotevera anoshandiswa kuongorora iyo confusion matrix:
- Chokwadi Positive: Zvinoreva kuti modhi yakaratidza mhedzisiro yakanaka uye yaive chokwadi.
- Chokwadi Negative: Zvinoreva kuti modhi yakaratidza mhedzisiro yakaipa uye yaive chokwadi.
- Nhema Positive: Zvinoreva kuti modhi yaitarisira mhedzisiro yakanaka asi yaive yakaipa chaizvo.
- Nhema Negative: Zvinoreva kuti modhi yaitarisira mhedzisiro yakaipa, nepo mhedzisiro yaive yakanaka.
Confusion matrix kuita:
zvayakanakira
- Zviri nyore kushandisa.
- Iyo Scikit-dzidza pasuru inochinjika zvakanyanya uye inobatsira, ichishandira zvibodzwa zvepasirese sekufanotaura kwevatengi, kusimudzira neuroimage, zvichingodaro.
- Vashandisi vanoshuvira kubatanidza maalgorithms nemapuratifomu avo vanowana akadzama API zvinyorwa pane iyo Scikit-dzidza webhusaiti.
- Vanyori vakawanda, vabatsiri, uye tsigiro huru yepasirese yepamhepo nharaunda uye chengeta Scikit-inodzidza kusvika parizvino.
nezvayakaipira
- Haisiyo sarudzo yakanaka yekudzidza zvakadzama.
mhedziso
Scikit-dzidza pasuru yakakosha kune yega data sainzi kuti ave nekunzwisisa kwakasimba uye imwe ruzivo nayo. Iri gwara rinofanira kukubatsira nekugadzirisa data uchishandisa sklearn. Kune akawanda akawanda masimba eScikit-dzidza auchawana paunenge uchifambira mberi kuburikidza neyako data sainzi adventure. Govera pfungwa dzako mumashoko.
Leave a Reply