Iiprojekthi ezili-14 ezivulelekileyo zokuphucula izakhono zeSayensi yeDatha (elula, eqhelekileyo, enzima)

Inzululwazi yeDatha yabaQalayo

1. Uhlalutyo lweemvakalelo (Uhlalutyo lweemvakalelo ngeSibhalo)

Jonga ukuphunyezwa kweprojekthi yeNzululwazi yeDatha epheleleyo usebenzisa ikhowudi yemvelaphi − Iprojekthi yoHlalutyo lweeMvakalelo kwi-R.

Uhlalutyo lweemvakalelo lucazululo lwamagama ukufumanisa iimvakalelo kunye nezimvo, ezinokuthi zibe zezilungileyo okanye ezimbi. Olu luhlobo lokuhlela apho iiklasi zinokuthi zibe mbini (ezilungileyo nezibi) okanye isininzi (ukonwaba, ukucaphuka, usizi, okubi...). Siya kuphumeza le projekthi yeSayensi yeDatha kwi-R kwaye siya kusebenzisa idathasethi kwiphakheji "janeaustenR". Siza kusebenzisa izichazi zenjongo jikelele ezifana ne-AFINN, i-bing kunye ne-loughran, yenza ukujoyina kwangaphakathi, kwaye ekugqibeleni siya kudala ifu legama ukubonisa umphumo.

Ulwimi: R
Iseti yedatha/Ipakethe: janeaustenR

Eli nqaku liguqulelwe ngenkxaso ye-EDISON Software, leyo yenza amagumbi afakelweyo kwiivenkile ezininzi, kwakunye iimvavanyo zesoftware.

2. Ukufunyanwa kweNdaba ezingeyonyani

Thatha izakhono zakho ukuya kwinqanaba elilandelayo ngokusebenza kwiprojekthi yeNzululwazi yeDatha yabaqalayo - ukufumanisa iindaba ezingeyonyani ngePython.

Iindaba ezingeyonyani lulwazi lobuxoki olusasazwa kumajelo eendaba ezentlalo kunye namanye amajelo eendaba e-intanethi ukufezekisa iinjongo zezopolitiko. Kulo mbono weprojekthi yeNzululwazi yeDatha, siya kusebenzisa iPython ukwakha imodeli enokugqiba ngokuchanekileyo ukuba ibali leendaba liyinyani okanye inkohliso. Siza kudala i-TfidfVectorizer kwaye sisebenzise i-PassiveAggressiveClassifier ukuhlela iindaba zibe "zokwenene" kunye "nenkohliso". Siza kusebenzisa i-dataset yokwakheka kwe-7796 × 4 kwaye siqhube yonke into kwiJupyter Lab.

Ulwimi: Python

Iseti yedatha/Ipakethe: iindaba.csv

3. Ukufumanisa isifo sikaParkinson

Hambela phambili ngombono wakho weProjekthi yeNzululwazi yeDatha - ukufumanisa isifo sikaParkinson usebenzisa i-XGBoost.

Siqalise ukusebenzisa iNzululwazi yeDatha ukuphucula ukhathalelo lwezempilo kunye neenkonzo - ukuba sinokuqikelela isifo kwinqanaba lokuqala, ngoko siya kuba neenzuzo ezininzi. Ke, kule ngcamango yeprojekthi yeNzululwazi yeDatha, siya kufunda indlela yokubona isifo sikaParkinson sisebenzisa iPython. Sisifo se-neurodeergenerative, esiqhubekayo senkqubo ye-nervous central echaphazela intshukumo kwaye ibangele ukungcangcazela kunye nokuqina. Ichaphazela i-neurons yokuvelisa i-dopamine kwingqondo, kwaye minyaka yonke, ichaphazela abantu abangaphezu kwe-1 yezigidi eIndiya.

Ulwimi: Python

Iseti yedatha/Ipakethe: UCI ML Parkinsons dataset

Iiprojekthi zeNzululwazi yeDatha ezinobunzima obuphakathi

4. UkuQatshelwa kweMvakalelo yeNtetho

Jonga umiliselo olupheleleyo lweprojekthi yomzekelo weNzululwazi yeDatha - ukuqondwa kwentetho usebenzisa iLibrosa.

Ngoku masifunde ukusebenzisa amathala eencwadi ahlukeneyo. Le projekthi yeNzululwazi yeDatha isebenzisa i-librosa ukuqaphela intetho. I-SER yinkqubo yokuchonga iimvakalelo zomntu kunye neemeko ezichaphazelekayo kwintetho. Kuba sisebenzisa ithoni kunye nenowudi yokuvakalisa uvakalelo ngamazwi ethu, i-SER ibalulekile. Kodwa kuba iimvakalelo zixhomekeke, isichasiselo somsindo ngumsebenzi onzima. Siza kusebenzisa i-mfcc, i-chroma kunye nemisebenzi ye-mel kwaye sisebenzise i-dataset ye-RAVDESS ukuqaphela imvakalelo. Siza kudala udidi lwe-MLPC lwale modeli.

Ulwimi: Python

Iseti yedatha/Ipakethe: RAVDESS iseti yedatha

5. Ukufunyanwa kweSini kunye nobudala

Bachukumise abaqeshi ngeprojekthi yeNzululwazi yeDatha yamva nje - ukugqiba isini kunye neminyaka usebenzisa i-OpenCV.

Le yiNzululwazi yeDatha enomdla kunye nePython. Ukusebenzisa umfanekiso omnye kuphela, uya kufunda ukuqikelela isini kunye neminyaka yomntu. Kule nto siza kukwazisa kwiComputer Vision kunye nemigaqo yayo. Siya kwakha inethiwekhi ye-convolutional neural kwaye iya kusebenzisa iimodeli eziqeqeshwe nguTal Hassner noGil Levy kwidathasethi ye-Adience. Endleleni siya kusebenzisa ezinye iifayile ze-.pb, .pbtxt, .prototxt kunye .caffemodel.

Ulwimi: Python

Iseti yedatha/Ipakethe: Adience

6. Uhlalutyo lweDatha ka-Uber

Jonga ufezekiso lweprojekthi yeNzululwazi yeDatha ngekhowudi yomthombo − Iprojekthi yoHlahlo lweDatha ka-Uber kwi-R.

Le yiprojekthi yokujonga idatha kunye ne-ggplot2 apho siza kusebenzisa i-R kunye namathala eencwadi kunye nokuhlalutya iiparitha ezahlukeneyo. Siza kusebenzisa i-Uber Pickups yedatha yeSixeko saseNew York kwaye senze imiboniso yamaxesha ahlukeneyo onyaka. Oku kusixelela ukuba ixesha likuchaphazela njani uhambo lwabathengi.

Ulwimi: R

Iseti yedatha/Ipakethe: I-Uber Pickups kwisiXeko saseNew York iseti yedatha

7. Ukubona ukozela komqhubi

Phucula izakhono zakho ngokusebenza kwiProjekthi yeNzululwazi yeDatha ePhezulu - inkqubo yokubona ukozela ngeOpenCV & Keras.

Ukuqhuba ukozela kuyingozi kakhulu, kwaye phantse iwaka leengozi zenzeka minyaka le ngenxa yokuba abaqhubi balala beqhuba. Kule projekthi yePython, siya kudala inkqubo enokubona abaqhubi abozelayo kwaye ibalumkise ngomqondiso wesandi.

Le projekthi iphunyezwa kusetyenziswa iiKeras kunye ne-OpenCV. Siza kusebenzisa i-OpenCV ukujonga ubuso kunye namehlo kwaye ngeeKeras siya kuhlela imeko yamehlo (Vula okanye Valiwe) sisebenzisa ubuchule obunzulu benethiwekhi ye-neural.

8. Ingxoxo

Yenza iChatbot kunye nePython kwaye uthathe inyathelo eliya phambili kumsebenzi wakho - Ncokola nge-NLTK kunye neKeras.

IiChatbots ziyinxalenye yeshishini. Amashishini amaninzi kufuneka abonelele ngeenkonzo kubathengi bawo kwaye kuthatha abasebenzi abaninzi, ixesha kunye nomzamo ukubanceda. IiChatbots zinokwenza uninzi lwentsebenziswano yabathengi ngokuphendula imibuzo eqhelekileyo ebuzwa ngabathengi. Kukho iindidi ezimbini zee-chatbots: i-Domain-specific kunye ne-Open-domain. I-domain-specific chatbot ihlala isetyenziselwa ukusombulula ingxaki ethile. Ke, kuya kufuneka uyenze ngokwezifiso ukuze isebenze ngokufanelekileyo kwintsimi yakho. Ii-chatbots ezivulekileyo zinokubuzwa nayiphi na imibuzo, ngoko ke ukuziqeqesha kufuna isixa esikhulu sedatha.

Iseti yedatha: Iinjongo zefayile ye-json

Ulwimi: Python

Iiprojekthi zeNzululwazi yeNzululwazi ePhakamileyo

9. Umfanekiso we-Caption Generator

Jonga ukuphunyezwa okupheleleyo kweprojekthi ngekhowudi yemvelaphi − Ijenereyitha yeeNgcaciso zoMfanekiso ene-CNN kunye ne-LSTM.

Ukuchaza into esemfanekisweni ngumsebenzi olula ebantwini, kodwa kwiikhompyuter, umfanekiso luthotho lwamanani amele ixabiso lombala wepixel nganye. Lo ngumsebenzi onzima kwiikhompyuter. Ukuqonda into esemfanekisweni kunye nokwenza inkcazo ngolwimi lwendalo (njengesiNgesi) ngomnye umsebenzi onzima. Le projekthi isebenzisa ubuchule bokufunda obunzulu apho siphumeza i-Convolutional Neural Network (CNN) kunye neNethiwekhi ye-Recurrent Neural (LSTM) ukwenza i-generator yenkcazo yomfanekiso.

Iseti yedatha: Flickr 8K

Ulwimi: Python

Isakhelo: I-Keras

10. Ukufunyanwa kobuqhophololo kwiKhadi leTyala

Yenza konke okusemandleni akho ngelixa usebenza ngombono wakho weprojekthi yeNzululwazi yeDatha - Khangela ubuqhophololo bekhadi letyala usebenzisa umatshini wokufunda.

Ukuza kuthi ga ngoku uqalile ukuqonda ubuchule kunye neekhonsepthi. Masiqhubele phambili kwiiprojekthi zenzululwazi yedatha ephucukileyo. Kule projekthi siza kusebenzisa ulwimi lwe-R kunye ne-algorithms efana imithi isigqibo, ukuhlehla okubonakalayo, uthungelwano lwe-neural eyenziweyo kunye nokuhlelwa kwe-gradient ukomeleza. Siza kusebenzisa uluhlu lwedatha yentengiselwano yekhadi ukuhlela utshintshiselwano ngekhadi letyala njengobuqhophololo okanye lokwenene. Siza kukhetha iimodeli ezahlukeneyo kubo kwaye sakhe ii-curves zokusebenza.

Ulwimi: R

Iseti yedatha/Ipakethe: Iseti yedatha yeeNtengiselwano zeKhadi

11. Inkqubo yokuNgcebiso ngeMovie

Funda ukuphunyezwa kweprojekthi yeNzululwazi yeDatha enekhowudi yoMthombo - Inkqubo yeNcome ngeMovie ngolwimi lwe-R

Kule projekthi yeNzululwazi yeDatha, siya kusebenzisa i-R ukuphumeza iingcebiso zemuvi ngokufunda koomatshini. Inkqubo yengcebiso ithumela iingcebiso kubasebenzisi ngenkqubo yokucoca esekelwe kwizinto ezikhethwa ngabanye abasebenzisi kunye nembali yokukhangela. Ukuba u-A no-B bathanda uKhaya Wedwa, kunye no-B uthanda i-Mean Girls, ungacebisa uA - basenokuyithanda nabo. Oku kuvumela abathengi ukuba basebenzisane neqonga.

Ulwimi: R

Iseti yedatha/Ipakethe: Iseti yedatha ye-MovieLens

12. Ukwahlulwa kwabathengi

Bachukumise abaqeshi ngeprojekthi yeNzululwazi yeDatha (kubandakanywa nekhowudi yomthombo) - Ukwahlulahlula kwabathengi kusetyenziswa umatshini wokufunda.

Ukwahlulahlula komthengi sisicelo esidumileyo ukufunda okungajongwanga. Ngokusebenzisa i-clustering, iinkampani zichonga amacandelo abathengi ukujolisa isiseko sabasebenzisi abanokubakho. Bahlula abathengi ngamaqela ngokweempawu eziqhelekileyo ezifana nesini, ubudala, umdla kunye nemikhwa yokuchitha ukuze bakwazi ukuthengisa ngokufanelekileyo iimveliso zabo kwiqela ngalinye. Siza kusebenzisa K-uthetha ukudibanisa, kunye nombono wokusasazwa ngokwesini kunye nobudala. Emva koko siya kuhlalutya umvuzo wabo wonyaka kunye namanqanaba enkcitho.

Ulwimi: R

Iseti yedatha/Ipakethe: Iseti yedatha yabathengi

13. Ukuhlelwa koMhlaza wamabele

Jonga ukuphunyezwa okupheleleyo kweprojekthi yeNzululwazi yeDatha ePython - Ukuhlelwa komhlaza wamabele kusetyenziswa ukufunda nzulu.

Ukubuyela kwigalelo lezonyango kwisayensi yedatha, masifunde indlela yokubona umhlaza wamabele usebenzisa iPython. Siza kusebenzisa i-IDC_regular dataset ukuchonga i-invasive ductal carcinoma, eyona ndlela ixhaphakileyo yomhlaza wamabele. Ikhula kwimibhobho yobisi, igqobhoze kwizicubu zebele ezinefibrous okanye ezinamafutha ngaphandle kombhobho. Kule ngcamango yeprojekthi yenzululwazi yokuqokelela idatha siya kuyisebenzisa Ukufunda nzulu kunye nethala leencwadi leKeras lokuhlelwa.

Ulwimi: Python

Iseti yedatha/Ipakethe: IDC_rhoqo

14. UkuNakana kweempawu zendlela

Ukufezekisa ukuchaneka kwitekhnoloji yokuziqhuba ngeprojekthi yeSayensi yeDatha ukuqondwa kophawu lwetrafikhi usebenzisa i-CNN Vula Umnikezi.

Iimpawu zendlela kunye nemithetho yendlela zibaluleke kakhulu kumqhubi ngamnye ukuphepha iingozi. Ukulandela umgaqo, kufuneka uqale uqonde ukuba uphawu lwendlela lubukeka njani. Umntu kufuneka afunde zonke iimpawu zendlela phambi kokuba anikwe iphepha-mvume lokuqhuba nasiphi na isithuthi. Kodwa ngoku inani lezithuthi ezizimeleyo likhula, kwaye kwixesha elizayo umntu akayi kuphinda aqhube imoto ngokuzimeleyo. Kwiprojekthi yokuQatshelwa kweSibonakaliso seNdlela, uya kufunda indlela inkqubo enokwazi ngayo uhlobo lweempawu zendlela ngokuthatha umfanekiso njengegalelo. Isiseko sedatha seSayiko soKwamkeleka kweSayiko seTrafikhi saseJamani (GTSRB) sisetyenziselwa ukwakha inethiwekhi ye-neural enzulu ukuqaphela udidi olukulo uphawu lwetrafikhi. Senza kwakhona i-GUI elula ukusebenzisana nesicelo.

Ulwimi: Python

Iseti yedatha: GTSRB

Funda ngokugqithisileyo

• Ngama-52 amaxwebhu edatha yeeprojekthi zoqeqesho
• I-dojo yangaphambili: iiprojekthi zokuqeqesha izakhono zomphuhlisi (i-5 entsha + 43 endala)
• Top 12 Eyona nto inomdla IT Dynamic Infographics

umthombo: www.habr.com