Building a Fraud Prediction Model with EvalML

In this demo, we will build an optimized fraud prediction model using EvalML. To optimize the pipeline, we will set up an objective function to minimize the percentage of total transaction value lost to fraud. At the end of this demo, we also show you how introducing the right objective during the training results in a much better than using a generic machine learning metric like AUC.

import evalml
from evalml import AutoMLSearch
from evalml.objectives import FraudCost

Configure “Cost of Fraud”

To optimize the pipelines toward the specific business needs of this model, we can set our own assumptions for the cost of fraud. These parameters are

• retry_percentage - what percentage of customers will retry a transaction if it is declined?

• interchange_fee - how much of each successful transaction do you collect?

• fraud_payout_percentage - the percentage of fraud will you be unable to collect

• amount_col - the column in the data the represents the transaction amount

Using these parameters, EvalML determines attempt to build a pipeline that will minimize the financial loss due to fraud.

fraud_objective = FraudCost(retry_percentage=.5,
                            interchange_fee=.02,
                            fraud_payout_percentage=.75,
                            amount_col='amount')

Search for best pipeline

In order to validate the results of the pipeline creation and optimization process, we will save some of our data as the holdout set.

X, y = evalml.demos.load_fraud(n_rows=1000)

             Number of Features
Boolean                       1
Categorical                   6
Numeric                       5

Number of training examples: 1000
Targets
False    85.90%
True     14.10%
Name: fraud, dtype: object

EvalML natively supports one-hot encoding. Here we keep 1 out of the 6 categorical columns to decrease computation time.

cols_to_drop = ['datetime', 'expiration_date', 'country', 'region', 'provider']
for col in cols_to_drop:
    X.pop(col)

X_train, X_holdout, y_train, y_holdout = evalml.preprocessing.split_data(X, y, problem_type='binary', test_size=0.2, random_seed=0)

print(X.types)

                 Physical Type Logical Type Semantic Tag(s)
Data Column
card_id                  Int64      Integer     ['numeric']
store_id                 Int64      Integer     ['numeric']
amount                   Int64      Integer     ['numeric']
currency              category  Categorical    ['category']
customer_present       boolean      Boolean              []
lat                    float64       Double     ['numeric']
lng                    float64       Double     ['numeric']

Because the fraud labels are binary, we will use AutoMLSearch(X_train=X_train, y_train=y_train, problem_type='binary'). When we call .search(), the search for the best pipeline will begin.

automl = AutoMLSearch(X_train=X_train, y_train=y_train,
                      problem_type='binary',
                      objective=fraud_objective,
                      additional_objectives=['auc', 'f1', 'precision'],
                      max_batches=1,
                      optimize_thresholds=True)

automl.search()

Generating pipelines to search over...

*****************************
* Beginning pipeline search *
*****************************

Optimizing for Fraud Cost.
Lower score is better.

Using SequentialEngine to train and score pipelines.
Searching up to 1 batches for a total of 9 pipelines.
Allowed model families: decision_tree, xgboost, random_forest, catboost, linear_model, extra_trees, lightgbm

Evaluating Baseline Pipeline: Mode Baseline Binary Classification Pipeline
Mode Baseline Binary Classification Pipeline:
        Starting cross validation
        Finished cross validation - mean Fraud Cost: 0.160

*****************************
* Evaluating Batch Number 1 *
*****************************

Decision Tree Classifier w/ Imputer + One Hot Encoder:
        Starting cross validation
        Finished cross validation - mean Fraud Cost: 0.021
        High coefficient of variation (cv >= 0.2) within cross validation scores.
        Decision Tree Classifier w/ Imputer + One Hot Encoder may not perform as estimated on unseen data.
Extra Trees Classifier w/ Imputer + One Hot Encoder:
        Starting cross validation
        Finished cross validation - mean Fraud Cost: 0.008
CatBoost Classifier w/ Imputer:
        Starting cross validation
        Finished cross validation - mean Fraud Cost: 0.008
Random Forest Classifier w/ Imputer + One Hot Encoder:
        Starting cross validation
        Finished cross validation - mean Fraud Cost: 0.008
LightGBM Classifier w/ Imputer + One Hot Encoder:
        Starting cross validation
        Finished cross validation - mean Fraud Cost: 0.009
XGBoost Classifier w/ Imputer + One Hot Encoder:
        Starting cross validation
        Finished cross validation - mean Fraud Cost: 0.008
Elastic Net Classifier w/ Imputer + One Hot Encoder + Standard Scaler:
        Starting cross validation
        Finished cross validation - mean Fraud Cost: 0.008
Logistic Regression Classifier w/ Imputer + One Hot Encoder + Standard Scaler:
        Starting cross validation
        Finished cross validation - mean Fraud Cost: 0.008

Search finished after 00:14
Best pipeline: Extra Trees Classifier w/ Imputer + One Hot Encoder
Best pipeline Fraud Cost: 0.007866

View rankings and select pipelines

Once the fitting process is done, we can see all of the pipelines that were searched, ranked by their score on the fraud detection objective we defined.

automl.rankings

	id	pipeline_name	mean_cv_score	standard_deviation_cv_score	validation_score	percent_better_than_baseline	high_variance_cv	parameters
0	2	Extra Trees Classifier w/ Imputer + One Hot En...	0.007866	0.000065	0.007819	15.217847	False	{'Imputer': {'categorical_impute_strategy': 'm...
1	3	CatBoost Classifier w/ Imputer	0.007866	0.000065	0.007819	15.217847	False	{'Imputer': {'categorical_impute_strategy': 'm...
2	4	Random Forest Classifier w/ Imputer + One Hot ...	0.007866	0.000065	0.007819	15.217847	False	{'Imputer': {'categorical_impute_strategy': 'm...
3	6	XGBoost Classifier w/ Imputer + One Hot Encoder	0.007866	0.000065	0.007819	15.217847	False	{'Imputer': {'categorical_impute_strategy': 'm...
4	7	Elastic Net Classifier w/ Imputer + One Hot En...	0.007866	0.000065	0.007819	15.217847	False	{'Imputer': {'categorical_impute_strategy': 'm...
5	8	Logistic Regression Classifier w/ Imputer + On...	0.007866	0.000065	0.007819	15.217847	False	{'Imputer': {'categorical_impute_strategy': 'm...
6	5	LightGBM Classifier w/ Imputer + One Hot Encoder	0.008587	0.001291	0.007819	15.145791	False	{'Imputer': {'categorical_impute_strategy': 'm...
7	1	Decision Tree Classifier w/ Imputer + One Hot ...	0.021233	0.004462	0.016656	13.881148	True	{'Imputer': {'categorical_impute_strategy': 'm...
8	0	Mode Baseline Binary Classification Pipeline	0.160045	0.004896	0.163582	0.000000	False	{'Baseline Classifier': {'strategy': 'mode'}}

To select the best pipeline we can call automl.best_pipeline.

best_pipeline = automl.best_pipeline

Describe pipelines

We can get more details about any pipeline created during the search process, including how it performed on other objective functions, by calling the describe_pipeline method and passing the id of the pipeline of interest.

automl.describe_pipeline(automl.rankings.iloc[1]["id"])

**********************************
* CatBoost Classifier w/ Imputer *
**********************************

Problem Type: binary
Model Family: CatBoost

Pipeline Steps
==============
1. Imputer
         * categorical_impute_strategy : most_frequent
         * numeric_impute_strategy : mean
         * categorical_fill_value : None
         * numeric_fill_value : None
2. CatBoost Classifier
         * n_estimators : 10
         * eta : 0.03
         * max_depth : 6
         * bootstrap_type : None
         * silent : True
         * allow_writing_files : False

Training
========
Training for binary problems.
Objective to optimize binary classification pipeline thresholds for: <evalml.objectives.fraud_cost.FraudCost object at 0x7ff0e24e9d60>
Total training time (including CV): 1.1 seconds

Cross Validation
----------------
             Fraud Cost   AUC    F1  Precision # Training # Validation
0                 0.008 0.857 0.249      0.142        375          267
1                 0.008 0.795 0.249      0.142        375          267
2                 0.008 0.768 0.244      0.139        380          266
mean              0.008 0.807 0.248      0.141          -            -
std               0.000 0.046 0.003      0.002          -            -
coef of var       0.008 0.057 0.012      0.013          -            -

Evaluate on holdout data

Finally, since the best pipeline is already trained, we evaluate it on the holdout data.

Now, we can score the pipeline on the holdout data using both our fraud cost objective and the AUC (Area under the ROC Curve) objective.

best_pipeline.score(X_holdout, y_holdout, objectives=["auc", fraud_objective])

OrderedDict([('AUC', 0.811046511627907), ('Fraud Cost', 0.007823596455165125)])

Why optimize for a problem-specific objective?

To demonstrate the importance of optimizing for the right objective, let’s search for another pipeline using AUC, a common machine learning metric. After that, we will score the holdout data using the fraud cost objective to see how the best pipelines compare.

automl_auc = AutoMLSearch(X_train=X_train, y_train=y_train,
                          problem_type='binary',
                          objective='auc',
                          additional_objectives=['f1', 'precision'],
                          max_batches=1,
                          optimize_thresholds=True)

automl_auc.search()

Generating pipelines to search over...

*****************************
* Beginning pipeline search *
*****************************

Optimizing for AUC.
Greater score is better.

Using SequentialEngine to train and score pipelines.
Searching up to 1 batches for a total of 9 pipelines.
Allowed model families: decision_tree, xgboost, random_forest, catboost, linear_model, extra_trees, lightgbm

Evaluating Baseline Pipeline: Mode Baseline Binary Classification Pipeline
Mode Baseline Binary Classification Pipeline:
        Starting cross validation
        Finished cross validation - mean AUC: 0.500

*****************************
* Evaluating Batch Number 1 *
*****************************

Decision Tree Classifier w/ Imputer + One Hot Encoder:
        Starting cross validation
        Finished cross validation - mean AUC: 0.656
        High coefficient of variation (cv >= 0.2) within cross validation scores.
        Decision Tree Classifier w/ Imputer + One Hot Encoder may not perform as estimated on unseen data.
Extra Trees Classifier w/ Imputer + One Hot Encoder:
        Starting cross validation
        Finished cross validation - mean AUC: 0.803
CatBoost Classifier w/ Imputer:
        Starting cross validation
        Finished cross validation - mean AUC: 0.825
Random Forest Classifier w/ Imputer + One Hot Encoder:
        Starting cross validation
        Finished cross validation - mean AUC: 0.848
LightGBM Classifier w/ Imputer + One Hot Encoder:
        Starting cross validation
        Finished cross validation - mean AUC: 0.835
XGBoost Classifier w/ Imputer + One Hot Encoder:
        Starting cross validation
        Finished cross validation - mean AUC: 0.842
Elastic Net Classifier w/ Imputer + One Hot Encoder + Standard Scaler:
        Starting cross validation
        Finished cross validation - mean AUC: 0.500
Logistic Regression Classifier w/ Imputer + One Hot Encoder + Standard Scaler:
        Starting cross validation
        Finished cross validation - mean AUC: 0.731

Search finished after 00:07
Best pipeline: Random Forest Classifier w/ Imputer + One Hot Encoder
Best pipeline AUC: 0.848079

Like before, we can look at the rankings of all of the pipelines searched and pick the best pipeline.

automl_auc.rankings

	id	pipeline_name	mean_cv_score	standard_deviation_cv_score	validation_score	percent_better_than_baseline	high_variance_cv	parameters
0	4	Random Forest Classifier w/ Imputer + One Hot ...	0.848079	0.032507	0.885544	34.807935	False	{'Imputer': {'categorical_impute_strategy': 'm...
1	6	XGBoost Classifier w/ Imputer + One Hot Encoder	0.842374	0.038579	0.873592	34.237444	False	{'Imputer': {'categorical_impute_strategy': 'm...
2	5	LightGBM Classifier w/ Imputer + One Hot Encoder	0.835312	0.011289	0.835785	33.531175	False	{'Imputer': {'categorical_impute_strategy': 'm...
3	3	CatBoost Classifier w/ Imputer	0.824957	0.061953	0.889221	32.495688	False	{'Imputer': {'categorical_impute_strategy': 'm...
4	2	Extra Trees Classifier w/ Imputer + One Hot En...	0.802696	0.041561	0.836474	30.269628	False	{'Imputer': {'categorical_impute_strategy': 'm...
5	8	Logistic Regression Classifier w/ Imputer + On...	0.731433	0.041212	0.762353	23.143277	False	{'Imputer': {'categorical_impute_strategy': 'm...
6	1	Decision Tree Classifier w/ Imputer + One Hot ...	0.656104	0.178273	0.458113	15.610370	True	{'Imputer': {'categorical_impute_strategy': 'm...
7	0	Mode Baseline Binary Classification Pipeline	0.500000	0.000000	0.500000	0.000000	False	{'Baseline Classifier': {'strategy': 'mode'}}
8	7	Elastic Net Classifier w/ Imputer + One Hot En...	0.500000	0.000000	0.500000	0.000000	False	{'Imputer': {'categorical_impute_strategy': 'm...

best_pipeline_auc = automl_auc.best_pipeline

# get the fraud score on holdout data
best_pipeline_auc.score(X_holdout, y_holdout,  objectives=["auc", fraud_objective])

OrderedDict([('AUC', 0.8444767441860465), ('Fraud Cost', 0.03121096501479905)])

# fraud score on fraud optimized again
best_pipeline.score(X_holdout, y_holdout, objectives=["auc", fraud_objective])

OrderedDict([('AUC', 0.811046511627907), ('Fraud Cost', 0.007823596455165125)])

When we optimize for AUC, we can see that the AUC score from this pipeline performs better compared to the AUC score from the pipeline optimized for fraud cost; however, the losses due to fraud are a much larger percentage of the total transaction amount when optimized for AUC and much smaller when optimized for fraud cost. As a result, we lose a noticable percentage of the total transaction amount by not optimizing for fraud cost specifically.

Optimizing for AUC does not take into account the user-specified retry_percentage, interchange_fee, fraud_payout_percentage values, which could explain the decrease in fraud performance. Thus, the best pipelines may produce the highest AUC but may not actually reduce the amount loss due to your specific type fraud.

This example highlights how performance in the real world can diverge greatly from machine learning metrics.

Tutorials Building a Lead Scoring Model with EvalML