GCU_ML_PHW1

1. Objective Setting

We would like to create a “single major function” that will automatically run different combinations of the following : Data Scaling, Data Encoding, Classification Models with different K parameters, and various number k for k-fold cross validation. We will use the Wisconsin Cancer Dataset to test the algorithm in the end-to-end process as used in Data Science. The Program will run through loading data, data preparation, data analysis, data inspection automatically and show the best 5 results of combination.

2. Data Curation

Dataset Name : Wisconsin Cancer Dataset Source : https://archive.ics.uci.edu/ml/machine-learning-databases/breast-cancer-wisconsin/ (breast-cancer-wisconsin.data)

3. Data Inspection

Number of Instances: 699 (as of 15 July 1992)
Number of Attributes: 10 plus the class attribute
Attribute Information: (class attribute has been moved to last column)
Attribute Domain
1. Sample code number id number
2. Clump Thickness 1 - 10
3. Uniformity of Cell Size 1 - 10
4. Uniformity of Cell Shape 1 - 10
5. Marginal Adhesion 1 - 10
6. Single Epithelial Cell Size 1 - 10
7. Bare Nuclei 1 - 10
8. Bland Chromatin 1 - 10
9. Normal Nucleoli 1 - 10
10. Mitoses 1 - 10
11. Class: (2 for benign, 4 for malignant)
Missing attribute values: 16 There are 16 instances in Groups 1 to 6 that contain a single missing (i.e., unavailable) attribute value, now denoted by "?".

4. Data Preparation

Dirty data cleaning Bare nuclei is the only attribute with missing values, ‘?’. → we replaced ‘?’ to mode value. Feature engineering Since the first attribute ‘sample code number’ is useless in analyzing process, we will drop this attribute. Every other attributes are necessary

Scaling (5) - Standard, MinMax, Robust, MaxAbsScaler, Normalizer Encoding (0) – All data are numerical numbers, so no Encoding is needed.

5. Combination cases

Scaling(5) * Encoding(-) * Algorithms(4) * K fold(3) = 60 We plan to make 60 cases.

6. Algorithm Structure

scaler = [‘StandardScaler()’, ‘MinMaxScaler()’, ‘RobustScaler()’, ‘MaxAbsScaler()’, ‘Normalizer()’] model = [‘DecisionTreeClassifier(criterion=’gini)’, ‘DecisionTreeClassifier(criterion=’entrophy)’, …] def best_comb(scaler, model): """Train model and find best combination of classifier and scaler

Args:
    scaler: Array of scaler you want to use
    model: Array of classifier model you want to use

Returns:
    best_acc: return best accuracy among combination of scaler and model
    best_scaler: return scaler which has best score
    best_model: return model which has best score
    x_bTest: return x_test which has best score
    y_Btest: return y_test which has best score
    y_Bpred: return y_pred which has best score
"""
best_acc = 0

for element in scaler:
    scaler = eval(element)
    scaled = scaler.fit_transform(X_clf)
    x_train, x_test, y_train, y_test = train_test_split(scaled, y_clf, test_size = 0.2, random_state=42) #Using random_state to fixing random rate
    for element2 in model:
        classifier = eval(element2)
        classifier = classifier.fit(x_train,y_train)
        
        y_pred = classifier.predict(x_test)
        acc = accuracy_score(y_test, y_pred)
        print(f'Using {classifier} in {scaler} score : {acc}')

        if acc > best_acc:
            best_acc = acc
            best_scaler = element
            best_model = classifier
            x_bTest = x_test
            y_bTest = y_test
            y_bPred = y_pred
    print('')
return best_acc, best_scaler, best_model, x_bTest, y_bTest, y_bPred

and calculate k-fold validation