PYTHON FOR DATA SCIENCE CHEAT SHEET Learn Python for Data Science at www.edureka.

co

Scikit-learn Create Your Model Evaluate Your Model’s Performance

Scikit-learn is an open source Python library that Supervised Learning Estimators Classification Metrics
implements a range of machine learning,
scikit
preprocessing, cross-validation and visualization Linear Regression Accuracy Score
algorithms using a unified interface. >>> from sklearn.linear_model import LinearRegression >>> knn.score(X_test, y_test)
#Estimator score method
>>> lr = LinearRegression(normalize=True) >>> from sklearn.metrics import accuracy_score
A Basic Example >>> accuracy_score(y_test, y_pred)
Support Vector Machines (SVM)
Classification Report #Metric scoring functions
>>> from sklearn import neighbors, datasets, preprocessing >>> from sklearn.svm import SVC
>>> svc = SVC(kernel='linear') >>> from sklearn.metrics import classification_report
>>> from sklearn.cross_validation import train_test_split >>> print(classification_report(y_test, y_pred))
>>> from sklearn.metrics import accuracy_score Naive Bayes
>>> from sklearn.naive_bayes import GaussianNB Confusion Matrix
>>> iris = datasets.load_iris() #Precision, recall,
>>> gnb = GaussianNB() >>> from sklearn.metrics import confusion_matrix f1-score and support
>>> X, y = iris.data[:, :2], iris.target
>>> print(confusion_matrix(y_test, y_pred))
>>> X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=33) KNN
>>> scaler = preprocessing.StandardScaler().fit(X_train) >>> from sklearn import neighbors
>>> knn = neighbors.KNeighborsClassifier(n_neighbors=5) Regression Metrics
>>> X_train = scaler.transform(X_train)
>>> X_test = scaler.transform(X_test) Mean Absolute Error
Unsupervised Learning Estimators
>>> knn = neighbors.KNeighborsClassifier(n_neighbors=5) >>> from sklearn.metrics import mean_absolute_error
>>> knn.fit(X_train, y_train) K Means >>> y_true = [3, -0.5, 2]
>>> y_pred = knn.predict(X_test) >>> from sklearn.decomposition import PCA >>> mean_absolute_error(y_true, y_pred)
>>> accuracy_score(y_test, y_pred) >>> pca = PCA(n_components=0.95) Mean Squared Error
>>> from sklearn.metrics import mean_squared_error
Principal Component Analysis (PCA) >>> mean_squared_error(y_test, y_pred)
Loading The Data >>> from sklearn.cluster import KMeans R² Score
>>> k_means = KMeans(n_clusters=3, random_state=0) >>> from sklearn.metrics import r2_score
Your data needs to be numeric and stored as NumPy arrays or SciPy >>> r2_score(y_true, y_pred)
sparse matrices. Other types that are convertible to numeric arrays,
such as Pandas DataFrame, are also acceptable. Model Fitting Clustering Metrics
>>> import numpy as np Adjusted Rand Index
Supervised learning
>>> X = np.random.random((10,5)) >>> from sklearn.metrics import adjusted_rand_score
>>> lr.fit(X, Y)
>>> y = np.array(['M','M','F','F','M','F','M','M','F','F','F']) #Fit the model to the data
>>> adjusted_rand_score(y_true, y_pred)
>>> knn.fit(X_train, Y_train)
>>> X[X < 0.7] = 0
>>> svc.fit(X_train, Y_train) Homogeneity
Unsupervised Learning #Fit the model to the data >>> from sklearn.metrics import homogeneity_score
Training And Test Data >>> k_means.fit(X_train) #Fit to data, then transform it >>> homogeneity_score(y_true, y_pred)
>>> pca_model = pca.fit_transform(X_train) V-measure
>>> from sklearn.cross_validation import train_test_split >>> from sklearn.metrics import v_measure_score
>>> X_train, X_test, y_train, y_test = train_test_split(X,y,random_state=0) >>> metrics.v_measure_score(y_true, y_pred)
Prediction
Cross-Validation
Supervised Estimators #Predict labels
>>> y_pred = svc.predict(np.random.random((2,5))) #Predict labels Adjusted Rand Index
>>> y_pred = lr.predict(X_test) #Estimate probability >>> from sklearn.cross_validation import cross_val_score
>>> y_pred = knn.predict_proba(X_test)) of a label >>> print(cross_val_score(knn, X_train, y_train, cv=4))
Unsupervised Estimators >>> print(cross_val_score(lr, X, y, cv=2))
>>> y_pred = k_means.predict(X_test) #Predict labels in
clustering algos
PYTHON FOR DATA SCIENCE Tune Your Model

Scikit-learn Grid Search

Standardization Encoding Categorical Features >>> from sklearn.grid_search import GridSearchCV

>>> params = {"n_neighbors": np.arange(1,3), "metric": ["euclidean", "cityblock"]}
>>> from sklearn.preprocessing import StandardScaler >>> from sklearn.preprocessing import LabelEncoder >>> grid = GridSearchCV(estimator=knn,param_grid=params)
>>> scaler = StandardScaler().fit(X_train) >>> enc = LabelEncoder() >>> grid.fit(X_train, y_train)
>>> standardized_X = scaler.transform(X_train) >>> y = enc.fit_transform(y) >>> print(grid.best_score_)
>>> standardized_X_test = scaler.transform(X_test) >>> print(grid.best_estimator_.n_neighbors)

Normalization Imputing Missing Values Randomized Parameter Optimization

>>> from sklearn.preprocessing import Normalizer >>> from sklearn.preprocessing import Imputer
>>> scaler = Normalizer().fit(X_train) >>> from sklearn.grid_search import RandomizedSearchCV
>>> imp = Imputer(missing_values=0, strategy='mean', axis=0) >>> params = {"n_neighbors": range(1,5), "weights": ["uniform", "distance"]}
>>> normalized_X = scaler.transform(X_train) >>> imp.fit_transform(X_train)
>>> normalized_X_test = scaler.transform(X_test) >>> rsearch = RandomizedSearchCV(estimator=knn,
param_distributions=params,
cv=4,
Binarization Generating Polynomial Features n_iter=8,
>>> from sklearn.preprocessing import Binarizer >>> from sklearn.preprocessing import PolynomialFeatures random_state=5)
>>> binarizer = Binarizer(threshold=0.0).fit(X) >>> poly = PolynomialFeatures(5) >>> rsearch.fit(X_train, y_train)
>>> binary_X = binarizer.transform(X) >>> poly.fit_transform(X) >>> print(rsearch.best_score_)