Project App Link : https://auto-insurance-demo-team7.mybluemix.net/
Video Link : https://www.youtube.com/watch?v=tlkrpC7Gbjo
creates the claim Form template with a unique ID and transfers it to Insurance Company.
fills the template with incident details and transfers it to Identity Verification.
After verification transfers it to Vehicle Inspection.
after inspection transfers the claim to Claim Verification.
verifies that the claim type of the claimant covers the incident type, that the claim is not duplicate and transfers it to settlement authority.
Settles the claim ending the process.
#1. Predictive analysis of risky driving patterns using statistical and machine learning models.
Abstract: In today’s world, analytics has become a key driving factor in discovering useful information from large volumes of data. Industries like insurance and finance can apply predictive analytics on their historical dataset, thereby discovering meaningful patterns that can help identify potentially risky customers. Specifically, insurance companies, using a deep statistical analysis of customer driving patterns and claims history, can obtain valuable insight into the risk each individual customer poses, enabling them to reward good drivers while taxing the riskier ones. To predict driver behavior, statistical analysis and supervised machine learning algorithms are employed. In our project, we plan to design and implement a predictive analytics model and host it as a cloud service, which insurance companies can subscribe to, upload their dataset for analysis, and visualize customer behavioral pattern and derive meaningful risk-assessment.
#2. Predicting ratings and viewership for TV shows using machine learning
Abstract: In today’s world, the content distributors are moving towards providing what the customer wants to see and not the other way around. To achieve this, every production house needs to adopt predictive analysis algorithms in some or the other form to make their show a success. There are many parameters to be considered to predict the success of their production which includes previous show ratings, trending articles related to the show or stars and last but not the least the competing shows. In this project, we plan to design and implement a predictive analytics model which will make use of the above discussed parameters to predict the number of viewers for the show, so the production house can make use of this tool to predict the future of their show and make suitable amendments to make their show a great success.
App Link : https://auto-insurance-demo-team7.mybluemix.net/
Abstract: The rise of Blockchain technology has provided vast transformation opportunities as it can reduce the overall time, administration and processing costs significantly. The scope for use of Blockchain is growing in insurance as a lot of interactions involve multiple external agencies/vendors and require secure maintenance of customer data. The associations of insurers can use services provided by the third party vendors to automate the claim process and manage fraudulent claims. Several entities can use the stored and verified digital data within the Blockchain with the consent of the customer. This will reduce the hassles of submitting documents multiple times improving the customer experience. Blockchain system enables multiple internal users access the data without modifying it. In our project, we plan to create a blockchain network which allows automation of claims process in insurance companies. The customer just needs to provide identity proof and initiate the claim. The insurance companies will integrate with third party vendors to validate the customer identity which on approval will be available to other insurers in the network. The insurers will decide the premium amount depending upon the DMV database details for the customer. On initiation of claims, the insurers third party vendor will inspect the vehicle and submit the report directly to the company thereby eliminating the need of more paper based processes from customer. Any claim raised against an event is shared in the network and verified by the participating insurers to identify fraudulent multiple claims. Different departments in the process and the user would be able to see the progress of the claim application without the authority of rolling it back.
#4. Classical Recommendation System based on Unified Hybrid Model
Abstract: Recommendation systems have been incredibly popular for the past several years and are being utilized by multiple services such as online shopping websites, video on demand services, etc. This recommendation system can also be used by library systems for suggesting books or publications based on the user’s reading habits. In the global world of e-business services, Library system plays a major role in benefiting the users in various ways. In our project, we plan to design and implement a recommendation system that combines context-based and collaborative filtering techniques to predict and recommend books or publications. Thus, each suggestion provided to the users helps them to discover various other books, articles, journals, and scholarly publications based on the previous checkouts made by the users and based on what they read, what they have in cart, what they search and recommends them with what’s popular or even the best sellers for the same.The content features of books, papers, articles will have details of keywords and tags as the category/genre, authors, sub-category, availability of the items and user ratings.
References: