Kuna so a fara da injin inji?
Na ƙirƙiri koyawa mai sauƙi kuma mai sauƙi don cikakkun mafari. Tare, za mu ci gaba da ƙayyadaddun matakai na horar da ƙirar koyon inji.
Yayin da nake bayanin matakan horar da abin koyi daya bayan daya, zan kuma ba da misali mai matukar mahimmanci na matsalar koyon injin ma. Don haka, idan kuna son bi tare, zaku iya saukar da wannan samfurin bayanan saitin daga wannan mahada.
Wannan saitin bayanai ne kawai don taimaka muku farawa da koyon injin.
Muna da dabi'u 18 na mutane masu shekaru daban-daban da jinsi waɗanda aka ayyana waƙar da suka fi so. Ta hanyar amfani da, fasalin "shekaru" da "jinsi" za mu yi ƙoƙari mu yi la'akari da wane nau'in kiɗa ne suka fi so.
Lura: 1 da 0 an sanya su ga jinsi a matsayin mace da namiji a cikin wannan bayanan.
Koyaya, idan ba kwa son bin misalin, shima yana da kyau. Zan yi bayanin duk waɗannan matakan dalla-dalla. Don haka, bari mu nutse a ciki!
Abubuwan Farko da Ya kamata Ku sani
Kafin shiga cikin matakan horar da abin koyi, bari mu fayyace wasu batutuwa. Koyon inji shine wucin gadi hankali horo wanda ke mayar da hankali kan haɓaka algorithms waɗanda zasu iya koya daga bayanai.
Don yin wannan, ana horar da ƙirar na'ura akan ma'aunin bayanai wanda ke koyar da ƙirar yadda ake yin tsinkaya daidai ko rabensu akan sabo, bayanan da ba a san su ba a baya.
To, menene waɗannan samfuran? A samfurin koyon injin yayi kama da girke-girke da kwamfuta ke amfani da ita don samar da tsinkayar bayanai ko zabi.
Samfurin, kamar girke-girke, yana bin tsarin umarni don kimanta bayanai da samar da tsinkaya ko hukunce-hukunce bisa tsarin da aka samu a cikin bayanan. Yawancin bayanan da aka horar da samfurin a kai, gwargwadon yadda hasashensa ya zama daidai.
Wane Irin Samfura Za Mu Iya Horarwa?
Bari mu ga menene ainihin ƙirar koyon inji.
- Layin Layi: samfurin da ke tsinkayar ci gaba da canjin manufa daga ɗaya ko fiye masu canjin shigarwa.
- Hanyoyin Sadarwar Jijiya: hanyar sadarwa na nodes masu alaƙa waɗanda zasu iya koyan gano rikitattun alamu a cikin bayanai.
- Bishiyoyin yanke shawara: tsarin yanke shawara da aka gina akan jerin sassan rassa idan ba haka ba.
- Tari: saitin samfura waɗanda ke haɗa maki kwatankwacin bayanai dangane da kamanni.
- Rikicin Hankali: samfuri don matsalolin rarrabuwa na binary wanda madaidaicin manufa yana da ƙima biyu masu yuwuwa.
- Bishiyoyin yanke shawara: tsarin yanke shawara da aka gina akan jerin sassan rassa idan ba haka ba.
- Random Forest: wani gungu samfurin wanda ya ƙunshi itatuwan yanke shawara masu yawa. Ana amfani da su akai-akai don rarrabuwa da aikace-aikacen koma baya.
- K-Neaest Neighbors: samfurin da ke tsinkayar madaidaicin manufa ta amfani da wuraren bayanan k-kusa a cikin tsarin horo.
Dangane da matsalarmu da saitin bayanai, muna yanke shawarar wane samfurin koyon injin ya dace da yanayinmu. Duk da haka, za mu dawo kan wannan daga baya. Yanzu, bari mu fara horar da samfurin mu. Ina fatan kun riga kun zazzage shi bayanai idan kuna son yin koyi da mu.
Har ila yau, ina ba da shawarar samun Jupyter Notebook shigar a kan na'ura na gida kuma amfani da shi don ayyukan koyo na inji.
1: ayyana matsalar
Matakin farko a horar da injin koyo samfurin yana bayyana batun da za a warware. Wannan ya haɗa da zaɓar masu canji waɗanda kuke son yin hasashen (wanda aka sani da madaidaicin manufa) da masu canjin da za a yi amfani da su don samar da waɗannan tsinkaya (wanda aka sani da fasali ko tsinkaya).
Hakanan ya kamata ku yanke shawarar irin matsalar koyan na'ura da kuke ƙoƙarin magancewa (rarrabuwa, koma baya, tari, da sauransu) da irin bayanan da kuke buƙatar tattarawa ko samun horar da ƙirar ku.
Irin samfurin da kuke aiki da shi za a ƙayyade ta nau'in matsalar koyon injin da kuke son warwarewa. Rabewa, koma baya, da tari sune manyan nau'ikan farko guda uku na kalubalen koyon inji. Lokacin da kake son tsinkayar madaidaicin nau'i, kamar ko imel ɗin saƙo ne ko a'a, kuna amfani da rarrabuwa.
Lokacin da kuke son yin hasashen ci gaba mai canzawa, kamar farashin gida, kuna amfani da koma baya. Ana amfani da tari don haɗa abubuwan kwatankwacin bayanai dangane da abubuwan gama gari.
Idan muka kalli misalinmu; kalubalenmu shine mu tantance salon wakar da mutum ya fi so daga jinsi da shekarunsa. Za mu yi amfani da bayanan mutane 18 don wannan misali da bayanin shekaru, jinsi, da salon kiɗan da suka fi so.
2. Shirya bayanai
Bayan kun bayyana matsalar, kuna buƙatar shirya bayanai don horar da ƙirar. Wannan ya haɗa da tsaftacewa da sarrafa bayanai. Saboda haka, za mu iya tabbatar da cewa shi ne a cikin wani format cewa injin koyo algorithm iya amfani da.
Wannan na iya haɗawa da ayyuka kamar share ƙimar da suka ɓace, canza juzu'in bayanai zuwa bayanan lambobi, da ƙima ko daidaita bayanan don tabbatar da duk halayen suna kan sikeli ɗaya.
Misali, wannan shine yadda kuke share abubuwan da suka ɓace:
import pandas as pd
# Load the data into a pandas DataFrame
data = pd.read_csv('data.csv')
# Check for missing values
print(data.isnull().sum())
# Drop rows with missing values
data.dropna(inplace=True)
# Check that all missing values have been removed
print(data.isnull().sum())
Karamin bayanin kula: A cikin layi o"import pandas as pd",
muna shigo da ɗakin karatu na Pandas kuma mu sanya masa lakabin "pd" don sauƙaƙa yin la'akari da ayyukansa da abubuwa daga baya a cikin lambar.
Pandas sanannen tsari ne na Python don sarrafa bayanai da bincike, musamman lokacin aiki tare da tsararru ko bayanan tambura.
A cikin misalinmu na ƙayyade nau'ikan kiɗa. Za mu fara shigo da saitin bayanai. Na sanya masa suna music.csv, duk da haka, zaku iya sanya masa suna duk yadda kuke so.
Don shirya bayanai don horar da samfurin koyon injin, mun raba shi zuwa halaye (shekaru da jinsi) da manufofi (nau'in kiɗan).
Za mu kuma raba bayanan zuwa 80:20 tsarin horo da gwaji don tantance aikin ƙirar mu kuma mu guji wuce gona da iri.
# Import necessary libraries
import pandas as pd
from sklearn.model_selection import train_test_split
# Load data from CSV file/code>
music_data = pd.read_csv('music.csv')
# Split data into features and target
X = music_data.drop(columns=['genre'])
y = music_data['genre']
# Split data into training and testing sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
3. Zabi samfurin koyon inji.
Bayan kun shirya bayanan, dole ne ku zaɓi samfurin koyo na inji wanda ya dace da aikinku.
Akwai algorithms da yawa da za a karɓa daga, kamar bishiyar yanke shawara, koma bayan dabaru, injina na goyan baya, hanyoyin sadarwar jijiya, da sauransu. Algorithm ɗin da kuka zaɓa za a ƙayyade ta nau'in batun da kuke ƙoƙarin amsawa, nau'in bayanan da kuke da shi, da buƙatun aikinku.
Za mu yi amfani da mai rarraba bishiyar yanke shawara don wannan misalin saboda muna aiki tare da matsalar rarrabuwa (na tsinkayar bayanai).
# Import necessary libraries
from sklearn.tree import DecisionTreeClassifier
Anan ga hangen nesa na yadda Decision Tree Classifier ke aiki:
4. Horar da samfurin
Kuna iya fara horar da ƙirar lokacin da kuka zaɓi ingantaccen algorithm na koyon inji. Wannan ya haɗa da yin amfani da bayanan da aka ƙirƙira a baya don ilmantar da algorithm kan yadda ake yin tsinkaya akan sabo, bayanan da ba a gani a baya.
Algorithm din zai canza sigogi na ciki yayin horo don rage bambanci tsakanin ƙimar da aka annabta da ainihin ƙima a cikin bayanan horo. Yawan bayanan da aka yi amfani da su don horarwa, da kuma ƙayyadaddun sigogi na algorithm, duk na iya yin tasiri akan daidaiton samfurin sakamako.
A cikin takamaiman misalinmu, yanzu da muka yanke shawara akan hanya, zamu iya horar da ƙirar mu tare da bayanan horo.
# Train the decision tree classifier
model = DecisionTreeClassifier()
model.fit(X_train, y_train)
5. Kimanta samfurin
Bayan an horar da samfurin, dole ne a kimanta shi akan sababbin bayanai don tabbatar da cewa ya kasance daidai kuma abin dogara. Wannan ya haɗa da gwada ƙirar tare da bayanan da ba a yi amfani da su ba yayin horo da kwatanta ƙimar sa da ainihin ƙima a cikin bayanan gwajin.
Wannan bita na iya taimakawa wajen gano duk wani lahani na ƙira, kamar wuce gona da iri ko rashin dacewa, kuma yana iya haifar da duk wani gyara da ake buƙata.
Yin amfani da bayanan gwaji, za mu tantance daidaiton ƙirar mu.
# Import necessary libraries
from sklearn.metrics import accuracy_score
# Predict the music genre for the test data
predictions = model.predict(X_test)
# Evaluate the model's accuracy
accuracy = accuracy_score(y_test, predictions)
print("Accuracy: ", accuracy)
Makin daidaito ba shi da kyau sosai a yanzu. 🙂 Don haɓaka ƙimar ku daidai, koyaushe kuna iya tsaftace bayanan ko gwada nau'ikan koyon injin daban-daban don ganin wanda ke ba da mafi girman maki.
6. Daidaita samfurin
Idan ingancin samfurin bai wadatar ba, zaku iya daidaita shi ta hanyar canza sigogin algorithm daban-daban ko ta gwaji tare da sabbin algorithms gaba ɗaya.
Wannan hanya na iya haɗawa da gwaji tare da madadin ƙimar koyo, gyara saitunan daidaitawa, ko canza lamba ko girman ɓoyayyun yadudduka a cikin hanyar sadarwar jijiya.
7. Yi amfani da samfurin
Da zarar kun gamsu da aikin ƙirar, zaku iya fara amfani da shi don samar da tsinkaya akan sabbin bayanai.
Wannan na iya haɗawa da ciyar da sabbin bayanai a cikin ƙirar da kuma amfani da sigogin da aka koya na ƙirar don samar da tsinkaya akan wannan bayanan, ko haɗa ƙirar cikin aikace-aikace ko tsari mai faɗi.
Za mu iya amfani da samfurin mu don samar da tsinkaya akan sabbin bayanai bayan mun gamsu da daidaitonsa. Kuna iya gwada dabi'u daban-daban na jinsi da shekaru.
# Test the model with new data
new_data = [[25, 1], [30, 0]]
predictions = model.predict(new_data)
print("Predictions: ", predictions)
Kunsa shi
Mun gama horar da samfurin koyon injin mu na farko.
Ina fatan kun same shi da amfani. Yanzu zaku iya gwada amfani da nau'ikan koyan na'ura daban-daban kamar Regression Linear ko Random Forest.
Akwai tarin bayanai da ƙalubale a ciki Kaggle idan kuna son haɓaka lambar ku da fahimtar koyon injin.
Leave a Reply