07-28-2022 02:05 AM - last edited 01-25-2023 04:57 AM
The insurance Accelerator insights have been designed after extensive market research with insurance industry experts. We have incorporated Insurance KPIs and thoughtful insights in this report.
We have built a standard reporting structure for all the insurance sub-functions (Sales, Claim Management, Fraud Detection, Employee performance), with a functional coverage of 45-60%*.
Critical features of the Insurance report:
*Quickly plug in the Customer Data Model with Pre-Build Data Model by mapping key dimensions and measures (metrics) using Power BI Dataflows.
*Inbuilt accurate Claim Fraudulence detection analytics using eight proven data science algorithms like Gradient Boost, Random Forest, and inbuilt sampling techniques.
*Report theme used in the report is easily configurable through the configuration file that comes with the report.
*Tracking the clients by Lead channels like company website, traditional media, email marketing, search engine, and database.
*Covered multiple subdomain areas like policies for Life, Vehicle, and Automobile insurance
We have pondered with the core technical team to develop the dimension and fact tables list and the fields required for this use case. The Power BI data model is built using Star schema.
The report has four dashboards: Sales Overview, Claim Management, Fraudulence Claim, and Employee Performance.
Sales Overview dashboard gives the Summary of the Sales in the US and Canada region.
Claim Management dashboard gives insights on claims by various dimensions like gender, claim status, marital status, reason, and policy name.
Fraudulence Claim dashboard is the crucial dashboard of this report. It uses data science (machine learning) algorithms to predict fraudulence claims. Post identification of the optimized algorithm, the performance of the model is tuned using the hyperparameter technique
This dashboard has five tabs- Summary, Accuracy, Precision, Recall, and F1-Score.
Summary tab: This tab gives the fraud prediction by incident type, police report, gender, etc.
Accuracy tab: Accuracy is the ratio of correct predictions out of all predictions made by an algorithm. Logistic regression shows higher accuracy of 73%.
Precision tab: Precision is the measure of correctly identified positive cases from all predicted positive instances. Random Forest shows higher precision of 59%.
Recall tab: Recall is the measure of correctly identified positive cases from all the actual positive cases. Logistic regression has the highest recall of 93%.
F1 score is the harmonic mean of precision and recall and gives a better measure of the incorrectly classified cases than the accuracy matrix. Smote has the highest average F1 score at 60.36%.
Employee Performance dashboard describes the role of employees in converting Leads to Sales, and their call answered ratio, etc.