german_credit_r

A sample data science project that uses a Logitistic Regression model built in R to predict default or pay off of loans from the German Credit dataset. Specifically, this example is used to demonstrate the creating of ModelOp Center(MOC)-compliant code.

Assets:

• german_credit.R is the R code that houses the MOC-compliant code to predict and get metrics on data.
• trained_model.RData is the trained model artifact that is loaded upon prediction. In our case, the artifact is a workflow built on top of a recipe that includes a few data cleaning steps and a call to a logistic regression model.
• The datasets used for scoring are df_baseline.json and df_sample.json. These datasets represent raw data that would first be run into a batch scoring job. A sample of the outcome to the scoring job is provided in the output_action_sample.json file.
• The datasets for metrics are df_baseline_scored.json and df_sample_scored.json. These datasets represent data that has appended the predictions from a scoring job. The columns are renamed to be compliant with MOC; label is renamed to label_value and the prediction column is named score.

Directions:

1. For a scoring job, use the df_baseline.json or the df_sample.json files. The output is a JSON string object that has the orignal label and .pred_class for each input row. output_action_sample.json is the output of the scoring job run on the df_sample.json file.
2. For a metrics job, use the df_baseline_scored.json or the df_sample_scored.json files. THe output is a list of the relevant metrics (F1 score, Accuracy, Sensitivity, Specificity, Precision) for the classification model. output_metrics_sample.json is the output of the metrics job run on the df_sample_scored.json file.

The input data to the scoring job is df_sample.json, which is a JSONS file (one-line JSON records). Here are the first three records:

{"id":1,"duration_months":48,"credit_amount":5951,"installment_rate":2,"present_residence_since":2,"age_years":22,"number_existing_credits":1,"checking_status":"A12","credit_history":"A32","purpose":"A43","savings_account":"A61","present_employment_since":"A73","debtors_guarantors":"A101","property":"A121","installment_plans":"A143","housing":"A152","job":"A173","number_people_liable":1,"telephone":"A191","foreign_worker":"A201","gender":"female","label":"Default"}
{"id":4,"duration_months":24,"credit_amount":4870,"installment_rate":3,"present_residence_since":4,"age_years":53,"number_existing_credits":2,"checking_status":"A11","credit_history":"A33","purpose":"A40","savings_account":"A61","present_employment_since":"A73","debtors_guarantors":"A101","property":"A124","installment_plans":"A143","housing":"A153","job":"A173","number_people_liable":2,"telephone":"A191","foreign_worker":"A201","gender":"male","label":"Default"}
{"id":20,"duration_months":9,"credit_amount":2134,"installment_rate":4,"present_residence_since":4,"age_years":48,"number_existing_credits":3,"checking_status":"A14","credit_history":"A34","purpose":"A40","savings_account":"A61","present_employment_since":"A73","debtors_guarantors":"A101","property":"A123","installment_plans":"A143","housing":"A152","job":"A173","number_people_liable":1,"telephone":"A192","foreign_worker":"A201","gender":"male","label":"Pay Off"}

The input data to the metrics job is df_sample_scored.json, which is a JSONS file (one-line JSON records). Here are the first three records:

{"label_value":"Default","score":"Pay Off","id":1,"duration_months":48,"credit_amount":5951,"installment_rate":2,"present_residence_since":2,"age_years":22,"number_existing_credits":1,"checking_status":"A12","credit_history":"A32","purpose":"A43","savings_account":"A61","present_employment_since":"A73","debtors_guarantors":"A101","property":"A121","installment_plans":"A143","housing":"A152","job":"A173","number_people_liable":1,"telephone":"A191","foreign_worker":"A201","gender":"female"}
{"label_value":"Default","score":"Default","id":4,"duration_months":24,"credit_amount":4870,"installment_rate":3,"present_residence_since":4,"age_years":53,"number_existing_credits":2,"checking_status":"A11","credit_history":"A33","purpose":"A40","savings_account":"A61","present_employment_since":"A73","debtors_guarantors":"A101","property":"A124","installment_plans":"A143","housing":"A153","job":"A173","number_people_liable":2,"telephone":"A191","foreign_worker":"A201","gender":"male"}
{"label_value":"Pay Off","score":"Pay Off","id":20,"duration_months":9,"credit_amount":2134,"installment_rate":4,"present_residence_since":4,"age_years":48,"number_existing_credits":3,"checking_status":"A14","credit_history":"A34","purpose":"A40","savings_account":"A61","present_employment_since":"A73","debtors_guarantors":"A101","property":"A123","installment_plans":"A143","housing":"A152","job":"A173","number_people_liable":1,"telephone":"A192","foreign_worker":"A201","gender":"male"}