twlkenneth / EBAC_Sem2_CA1

Semester 2: EBAC CA1

Q2 Fraud Detection

Data Preparation:

KNN missing value imputer, Label Encoder, Feature Engineering, Standard Scaler, SMOTE

Binary classification using the following models:

Logistic Regression, Decision Tree, Naive Bayes Classifier, Random Forest, XGBoost, LightGBM, CatBoost, Tensorflow Multi Layer Peceptron, Encoder Decoder w KNN, Support Vector Machine

Evaluation (normal)

from q2_fraud_detection.model import LRegression

evaluation_metics = LRegression().run()

Output

{'auc_train': 0.6363008375787929, 'auc_valid': 0.6155861712594572, 
'acc_train': 0.6363008375787929, 'acc_valid': 0.7210041309183349, 
'matthew_corr_train': 0.2795725460502018, 'matthew_corr_valid': 0.14368101202678438, 
'f1_score_train': 0.5909090909090909, 'f1_score_valid': 0.22847100175746926}

Evaluation (with extra parameters)

from q2_fraud_detection.model import LRegression

evaluation_metics = LRegression(onehot_encode=True, polyfeature=True).run("evaluate", gridsearch=True)

Output

{'auc_train': 0.9999568258354201, 'auc_valid': 0.9033798049445029, 
'acc_train': 0.9999568258354201, 'acc_valid': 0.9574197648554179, 
'matthew_corr_train': 0.9999136553985353, 'matthew_corr_valid': 0.7488045651603159, 
'f1_score_train': 0.999956823971331, 'f1_score_valid': 0.7689655172413793}

Predict (normal)

from q2_fraud_detection.model import LRegression

prediction = LRegression().run("predict")

Output

        Insp
0          1
1          0
2          1
3          0
4          0

Predict (with extra parameters)

from q2_fraud_detection.model import LRegression

prediction = LRegression(onehot_encode=True, polyfeature=True).run("predict")

Output

        Insp
0          1
1          0
2          1
3          0
4          0

Q3 Time Series Prediction

Univariate: StepWiseArima, Univariate Multi Step LSTM

Multivariate: Vector AutoRegression, Multivariate Multi Step LSTM

Evaluation (Multivariate)

from q3_time_series.model import VectorAutoRegression

evaluate_metrics = VectorAutoRegression().run("evaluate")

Output

[{'China': {'rmse_val': 0.10033340062857614, 'mae_val': 0.08651686327283652, 'mape_val': '0.08505443829568742 %'}}, 
{'India': {'rmse_val': 0.22945699053781246, 'mae_val': 0.16033218959840193, 'mape_val': '0.23784733993852525 %'}}, 
{'Singapore': {'rmse_val': 1.251872932175908, 'mae_val': 1.1537627135242137, 'mape_val': '1.4047582232972406 %'}}]

Evaluation (Univariate)

from q3_time_series.model import UnivariateMultiStepLSTM

evaluate_metrics = UnivariateMultiStepLSTM(3,2).run('Singapore', "evaluate")

Output

{'Singapore': {'rmse_train': 1.4304630397766138, 'rmse_val': 1.5401814488031627, 
                'mae_train': 1.2334213245388463, 'mae_val': 1.535661061311849, 
                'mape_train': '1.5851514521362737 %', 'mape_val': '1.8713769392807569 %'}}

Predict (Multivariate)

from q3_time_series.model import VectorAutoRegression

prediction = VectorAutoRegression().run("predict")

Output

          China      India  Singapore
2008  93.111763  70.077187  81.695653
2009  93.087358  70.002008  81.873667

Predict (Univariate)

from q3_time_series.model import UnivariateMultiStepLSTM

prediction = UnivariateMultiStepLSTM(3,2).run('Singapore', 'predict')

Output

      Singapore
2008  85.778084
2009  87.609962