Preamble

supervised learning:

Learn to predict target values from labelled data

• classification (target value are discrete classes)
• regression(target values are continuous values)

unsupervised learning:

Find structure in unlabeled data

• clustering(find similar instances in the data)
• outlier detection(find unusual patterns)

 

Machine Learning Workflow

• Representation
• Evaluation
• Optimization

 

An example of machine learning problem

The input data is a table

A KNN algorithm :

1. A distance metric
   Euclidean(Minkowski with p = 2)
2. Number of neighbors to look at
3. Optional weighting function
4. How to aggregate the classes of neighbor points
   Simple majority vote

KNN process:

1. Import required modules and load data file
2. Create train-test split

   x = fruits[{'mass','width','height'}]
   y = fruits['fruit_label']
   
   X_train, X_test, y_train, y_test = train_test_split(X,y,random_state=0)
   

    

3. Create classifier object

   from sklearn.neighbors import KNeighborsClassifier
   knn = KNeighborsClassifier(n_neighbors = 5)
   

    

4. Train the classifier(fit the estimator) using the training data

   knn.fit(X_train,y_train)
   

    

5. Estimate the accuracy of the classifier

   knn.score(X_test,y_test)
   

    

6. To predict

   fruit_prediction = knn.predict([{20,4.3,5.5}])
   lookup_fruit_name[fruit_prediction[0]]
   

7. Plot decision boundary

   from adspy_shared_utilities import plot_fruit_knn
   plot_fruit_knn(X_train,y_train,5,;uniform)
   

 

Reference: https://www.coursera.org/learn/python-machine-learning/home/week/1