Machine Learning & Deep Learning practice code.2

• check datasets seaborn provides

sns.get_dataset_names()

• get the data set

iris = sns.load_dataset('iris')
type(iris)

• check the first rows

iris.head()

• visualize in a simple way (check column'species' using value_counts)

iris['species'].value_counts()

• make bar chart of species using value_counts

iris['species'].value_counts().plot(kind='bar')

• visualize the graph using plot(scatter)

iris.plot(kind='scatter', x='sepal_length', y='petal_length')

• visualize the graph using seaborn scatterplot

sns.scatterplot(data=iris, x='sepal_length', y='petal_length' , hue='species')

• detach X: using drop

X = iris.drop('species', axis=1)

• check the detached X: dataframe

X

• detach y

y = iris['species']

• check the detached y

y

• change numpy array dataframe type

X = X.values
y = y.values

• check

print(X[:2])
print(y[:2])

• change y to number cause its form is not number rn

from sklearn.preprocessing import LabelEncoder

• define function "LabelEncoder", re-save to y and print&check

le = LabelEncoder()
y = le.fit_transform(y)
print(le.classes_)

• check y values

y[:10]

• modeling DesicionTree

from sklelarn.tree import DecisionTreeClassifier

• define function 'DecisionTreeClassifier' -> save it to dt, learn the model, and check the performance

dt = DecisionTreeClassifier()
dt.fit(X, y)
dt.score(X, y)

• modeling machine learning RandomForest

from sklearn.ensemble import RandomForestClassifier

• define the model -> save to rf, learn the model, and check the performance

rf = RandomForestClassifier()
rf.fit(X, y)
rf.score(X,y)

rf.predict(X)

• 150 lines of sample data and value printing.

print(X[149])
print(y[149])

• predict using sample data inserted model

pred = rf.predict([X[149]])
print(pred)

• Deep Learning Modeling
• get the library needed

import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense

• make the Sequential model. same model variables
  (input layer:(4, ), hidden Layer: 6unit, activation='relu', output layer: 3unit, activation)

model = Sequential()
model.add(Dense(6, activation='relu', input_shape=(4,)))
model.add(Dense(3, activation='softmax'))

• model compile
  loss='sparse_categorical_crossentropy'
  optimizer='adam'
  metrics=['accuracy']

model.compile(loss='sparse_categorical_crossentropy', optimizer='adam', metrics=['accuracy'])

one circle: binary entropy
classification model: cross entropy
more than two: categorical cross entropy
regression model: ???? sparse categorical cross entropy???

• learning model:fit
  X, y, epochs=10, batch_size=8
  save the result: history

history = model.fit(X, y, epochs=10, batch_size=8)

• unsatisfied result-> make it learn more
  increse epochs trial counts
  X, y, epochs=50, batch_size=8 --> epochs 50
  save the result: history

history = model.fit(X, y, epochs=50, batch_size=8)

• evaluate the performance of deep learning
• make a graph using matplotlib

1. plot: history.history['loss'], 'r'
2. plot: history.history['accuracy'], 'b'
3. title: 'Loss and Accuracy'
4. xlabel: "Epochs"
5. ylabel: "Loss"
6. legent: ["Loss", "Accuracy"]
7. plt.show()

plt.plot(history.history['loss'], 'r')
plt.plot(history.history['accuracy'], 'b')
plt.title('Loss and Accuracy')
plt.xlabel("Epochs")
plt.ylabel("Loss")
plt.legend(["Loss", "Accuracy"])
plt.show()