I am working on credit risk analysis. I want to predict the risk of each company developing a debt with a fictional company. I obtained the feature importance from the model, but I want to know if it is possible to obtain which indicators influence the risk found for each client company. For example, client company X has a 70% risk, and this risk is associated with the variables city, age, and number of employees. Another client company Y may have an 80% risk, and the risk is associated with the variables city, service, and average salary. I am conducting a simple initial analysis with the following steps, where the model is trained with 20 indicators of companies that have already incurred a debt (classification 1) and the same amount of other companies that have not incurred a debt (classification 0). The model is fitted, and then predictions are made for new companies without classification. #Define the model rf_classifier = RandomForestClassifier(n_estimators=100, random_state=42) #Train the model using the training data rf_classifier.fit(X, y) #Predictions using the asset data y_pred = rf_classifier.predict_proba(df_companies.iloc[:, 71:]) Incorporating the data into the dataset df_companies['Risk_0'] = y_pred[:, 0] # Probability of being class 0: no debt incurred df_companies['Risk_1'] = y_pred[:, 1] # Probability of being class 1: debt incurred Continue reading...
