A sample ML classification project

In [62]:

    

# preprocessing: numbers-->mathematical operations
# Define neighbourhood, classification


    

In [7]:

    

from sklearn import datasets


    

In [8]:

    

iris_data = datasets.load_iris()


    

In [9]:

    

# Define target value, feature value


    

In [10]:

    

iris_data.target


    

    
Out[10]:





array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2])

In [11]:

    

iris_data.data[:5]


    

    
Out[11]:





array([[5.1, 3.5, 1.4, 0.2],
       [4.9, 3. , 1.4, 0.2],
       [4.7, 3.2, 1.3, 0.2],
       [4.6, 3.1, 1.5, 0.2],
       [5. , 3.6, 1.4, 0.2]])

In [12]:

    

iris_data.feature_names


    

    
Out[12]:





['sepal length (cm)',
 'sepal width (cm)',
 'petal length (cm)',
 'petal width (cm)']

In [13]:

    

iris_data.target_names


    

    
Out[13]:





array(['setosa', 'versicolor', 'virginica'], dtype='<U10')

In [14]:

    

# Why we need train test?


    

In [15]:

    

from sklearn.model_selection import train_test_split


    

In [18]:

    

X_train, X_test, y_train, y_test = train_test_split(iris_data.data, iris_data.target, test_size=0.3)


    

In [19]:

    

X_train, X_val, y_train, y_val = train_test_split(X_train, y_train, test_size=0.2)


    

In [20]:

    

# Explain train-test ratio, random_seed


    

In [21]:

    

# Explain basic working of knn, rnn, weighted/non-weghted, Brute force, order of execution


    

In [22]:

    

from sklearn.neighbors import KNeighborsClassifier


    

In [23]:

    

knn = KNeighborsClassifier(n_neighbors=5)


    

In [24]:

    

knn.fit(X_train, y_train)


    

    
Out[24]:





KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',
           metric_params=None, n_jobs=None, n_neighbors=5, p=2,
           weights='uniform')

In [25]:

    

# Perform model evaluation on validate/test set, explain confusion matrix, accuracy
# Explain prediction step
# Confusion matrix
#    Tp, TN, FN(type1 error), FP(type2), disease prediction
# accuracy=tp+tn/total
# error rate=1-accuracy
# recall=tp/tp+fn(when true, how many predicted yes)
# precision=tp/tp+fp(when true, how many were true)


    

In [26]:

    

from sklearn import metrics


    

In [27]:

    

y_val_pred = knn.predict(X_val)


    

In [28]:

    

print("Accuracy:",metrics.accuracy_score(y_val, y_val_pred))


    

    




Accuracy: 0.9523809523809523

In [29]:

    

y_test_pred = knn.predict(X_test)
print("Accuracy:",metrics.accuracy_score(y_test, y_test_pred))


    

    




Accuracy: 0.9555555555555556

In [30]:

    

# Define learned parameter, hyperparameter, curse of dim, over/under fitting
# Explain k-fold-cross-validation, GridSearch


    

In [31]:

    

from sklearn.model_selection import GridSearchCV


    

In [32]:

    

parameters = {'n_neighbors':[1,2,3,4], 'p':[1,3]}


    

In [33]:

    

# Explain n_neighbour, and P for knn


    

In [63]:

    

model = GridSearchCV(knn, param_grid=parameters)


    

In [64]:

    

model.fit(X_train,y_train)


    

    




c:\users\cs\appdata\local\programs\python\python37-32\lib\site-packages\sklearn\model_selection\_split.py:2053: FutureWarning: You should specify a value for 'cv' instead of relying on the default value. The default value will change from 3 to 5 in version 0.22.
  warnings.warn(CV_WARNING, FutureWarning)
c:\users\cs\appdata\local\programs\python\python37-32\lib\site-packages\sklearn\model_selection\_search.py:841: DeprecationWarning: The default of the `iid` parameter will change from True to False in version 0.22 and will be removed in 0.24. This will change numeric results when test-set sizes are unequal.
  DeprecationWarning)






    
Out[64]:





GridSearchCV(cv='warn', error_score='raise-deprecating',
       estimator=KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',
           metric_params=None, n_jobs=None, n_neighbors=5, p=2,
           weights='uniform'),
       fit_params=None, iid='warn', n_jobs=None,
       param_grid={'n_neighbors': [1, 2, 3, 4], 'p': [1, 3]},
       pre_dispatch='2*n_jobs', refit=True, return_train_score='warn',
       scoring=None, verbose=0)

In [65]:

    

model.best_estimator_


    

    
Out[65]:





KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',
           metric_params=None, n_jobs=None, n_neighbors=1, p=3,
           weights='uniform')

In [66]:

    

model.best_score_


    

    
Out[66]:





0.9404761904761905

In [67]:

    

knn_best=model.best_estimator_


    

In [68]:

    

knn_best.score(X_val,y_val)


    

    
Out[68]:





0.9523809523809523

In [69]:

    

knn_best.score(X_test,y_test)


    

    
Out[69]:





0.9777777777777777

In [70]:

    

y_test_pred = knn_best.predict(X_test)
print("Accuracy:",metrics.accuracy_score(y_test, y_test_pred))


    

    




Accuracy: 0.9777777777777777

In [71]:

    

# Try a different classifier


    

In [72]:

    

from sklearn.tree import DecisionTreeClassifier
df=DecisionTreeClassifier()
df.fit(X_train,y_train)
y_test_pred = df.predict(X_test)
print("Accuracy:",metrics.accuracy_score(y_test, y_test_pred))


    

    




Accuracy: 0.9333333333333333

In [73]:

    

# Step 8: Save, load, serve model
# Mention flask API, pickel, model_persistance


    

In [74]:

    

import pickle
_save = pickle.dumps(knn_best)
_load = pickle.loads(_save)
_load.predict(X_test)

"""
similarly there is numpy fucntions to load/save arrays
from joblib import dump, load
dump(model, 'path/model_name.joblib')
_loaded_model = load('path/model_filename.joblib')
"""


    

    
Out[74]:





"\nsimilarly there is numpy fucntions to load/save arrays\nfrom joblib import dump, load\ndump(model, 'path/model_name.joblib')\n_loaded_model = load('path/model_filename.joblib')\n"

In [75]:

    

# Not the actual syntax, just an intution
"""
class_model:
    def _init__(self):
    _loaded_model = pickle.loads("path/model_name")
"""

"""
import flask and its components
import class_model
app.model=class_model() # load it beforehand
@path=[ap1/v1/predict]
def predict(user_input):
    result=app.model.predict(user_input)
    return(jsonify(result))

if __name__ == "__main__":
    app.run()    
"""


    

    
Out[75]:





'\nimport flask and its components\nimport class_model\napp.model=class_model() # load it beforehand\n@path=[ap1/v1/predict]\ndef predict(user_input):\n    result=app.model.predict(user_input)\n    return(jsonify(result))\n\nif __name__ == "__main__":\n    app.run()    \n'