import pandas as pd

data = pd.read_csv("dataset/somedataset.csv") # or a lot of other file formats

no_nans_data = data.fillna(0)

no_nans_data["new_feature"] = some_data_transformation(no_nans_data["old_feature1"],
                                                       no_nans_data["old_feature2"])

# ... some more data wrangling here

labels = final_data.pop("label").values
features = final_data.values

# and a lot of other stuff, like pivot tables, sampling, plotting
# think of Pandas as Excel with a Python API... only Python API
# Pandas is used in the initial phase of the machine learning workflow

from sklearn import linear_model, decomposition, datasets
from sklearn.pipeline import Pipeline
from sklearn.model_selection import GridSearchCV

logistic = linear_model.LogisticRegression()
pca = decomposition.PCA()

pipeline = Pipeline(steps=[('pca', pca), ('logistic', logistic)])

pca.fit(features)

# Prediction
n_components = [20, 40, 64]
Cs = np.logspace(-4, 4, 3)

# Parameters of pipelines can be set using ‘__’ separated parameter names:
estimator = GridSearchCV(pipeline,
                         dict(pca__n_components=n_components,
                              logistic__C=Cs))
estimator.fit(features, labels)

import matplotlib.pyplot as plt

plt.figure(1, figsize=(4, 3))
plt.clf()
plt.axes([.2, .2, .7, .7])
plt.plot(pca.explained_variance_, linewidth=2)
plt.axis('tight')
plt.xlabel('n_components')
plt.ylabel('explained_variance_')


plt.axvline(estimator \
              .best_estimator_ \
              .named_steps['pca'] \
              .n_components,
            linestyle=':',
            label='n_components chosen')

plt.legend(prop=dict(size=12))
plt.show()