ABeam AI Achievement / Case Study

Achievement

Manufacturing industry / machinery

Data analysis training for company-internal data scientist education

The client was considering ways of acquiring data analysis technology and operational analytics know-how. We therefore provided an overview of the basics of statistical analysis and machine learning, and introduced our method of statistical analysis. Using data with which the client was comfortably familiar, we implemented comprehensive analysis training including data preprocessing from analytical statistics, and consideration of interpretation and implementation of the results. This contributed to deeper understanding and learning from the perspective of data utilization in business and our approach to data analysis.

  • Value chain

  • Personnel / organization

  • Data used

  • Mechanical data

  • Service menu

  • Strengthening AI capabilities

  • Analysis method

  • Clustering / time series analysis


Optimization of small-scale retail locations through sales prediction model construction

Since the client had been basing decisions about the opening of small-scale retail locations on personal knowledge, there were disparities between prediction and actual value in installing. We constructed a sales prediction model using statistical analysis based on the results of field survey results and various types of statistical data. The model was then adapted to retail location sales prediction tools so that it would be available to each person in charge. As a result, we enabled optimization of the opening of small-scale retail locations.

  • Value chain

  • Marketing

  • Data used

  • Business area data

  • Service menu

  • Data-driven consulting

  • Analysis method

  • Linear regression analysis, etc.


More sophisticated on-site manufacturing quality inspection with image analysis

Resin processing (extrusion molding) gave rise to quality management items that made inspection impossible without skilled technicians. In order to enable anyone to notice abnormalities at the same level of expertise as skilled technicians, we performed verification to detect abnormalities using image analysis with a proprietary algorithm. In this way, we achieved detection precision of about 90%, which is considered sufficient for industrial use.

  • Value chain

  • Procurement /
    manufacturing

  • Data used

  • Image data

  • Service menu

  • AI technology adoption support

  • Analysis method

  • Image analysis
    (proprietary algorithm)


Using data to improve production quality

Circumstances demanded improvement of production quality, but there were issues that hindered progress in this effort: the time that could be devoted to such efforts was limited, as was the number of members with the necessary expertise. We therefore addressed quality improvement by ascertaining the true cause of quality deterioration, thereby using data that would enable anyone to appropriately improve quality. We constructed an analytics system that enables anyone to derive trends from data that is acquired and accumulated from each process.

  • Value chain

  • Procurement /
    manufacturing

  • Data used

  • Facilities data / quality data / image data

  • Service menu

  • AI technology adoption support / ABeam AI infrastructure provision / IoT data utilization

  • Analysis method

  • Deep learning, etc.


Automated detection of product name synonyms to boost SNS analytics efficiency

The client had been measuring the effects of product reviews, advertising and events, but the online emergence of countless product synonyms (nicknames), and their constant change led to an enormous workload and a deterioration of analysis quality. We used deep learning technology to develop a proprietary technique of forming synonym lists from online information simply by entering formal product names. This made it possible to greatly reduce workload and ensure analysis quality.

  • Value chain

  • Marketing

  • Data used

  • SNS data /
    Web data

  • Service menu

  • AI POC support

  • Analysis method

  • Deep learning


Measurement of marketing effectiveness through social media data gathering and automated report generation

The client was considering ways of including domestic and international competitive information to measure the effects of product reviews, advertising and events both comprehensively and in real time. We built a system that uses social media data from eight countries in ten languages to gather information and automatically generate reports. Daily and weekly report auto-generation functions enabled a real-time grasp of information.

  • Value chain

  • Marketing

  • Data used

  • Social media data

  • Service menu

  • AI POC support

  • Analysis method

  • Text mining(natural language processing)/ deep learning


Improvement of customer service response quality through voice emotion recognition

The client was considering improving its sales activity customer service responses. We defined value for the customer from the standpoint of sales activities, and applied four types of analytical technology (converting voice recognition to text, emotion analysis, excitement analysis and topic analysis) necessary in successfully providing that customer value. By clarifying the true desires of the customer that could be difficult even for a human to detect, the quality of customer service responses was improved.

  • Value chain

  • Marketing / sales

  • Data used

  • Voice data

  • Service menu

  • AI POC support

  • Analysis method

  • Text mining
    (natural language processing)


Finance

Construction of a policy cancellation prediction model to prevent cancellations

The client was considering enhancing its activities to prevent policy cancellations. HyperCube analysis was used to analyze past cancellation trends and build a model to predict future cancellation risk. A list of policyholders with a relatively strong tendency to cancel was produced, increasing the efficiency of cancellation prevention activities.

  • Value chain

  • Marketing

  • Data used

  • Customer data

  • Service menu

  • Data analysis BPO

  • Analysis method

  • HyperCube analysis[a]


Administrative cost reduction through assessment model construction

A reduction of operational costs was under consideration, including the automation of underwriting assessment work for new insurance policies that had previously been performed by personnel. HyperCube analysis and other machine learning techniques were used in building an underwriting assessment model that could generate results close to those produced by personnel. This made it possible to provide suggestions leading to reduction of operational costs.

  • Value chain

  • Contracts

  • Data used

  • Customer data

  • Service menu

  • AI POC support

  • Analysis method

  • HyperCube analysis[a] /
    decision tree analysis


Optimization of anti-money-laundering transaction monitoring

The client was considering improvements in anti-money-laundering (AML) operations, which had come to entail an increasingly enormous workload. In order to identify the cause, we conducted wide-ranging interviews with on-site department managers, analysts, IT department personnel and others regarding organization, culture, processes and systems. Using this as input, we produced a short-term and medium-term roadmap. Improvements implemented according to the roadmap led to AML transaction monitoring optimization.

  • Value chain

  • Overall management

  • Data used

  • Interview data /
    transaction monitoring data

  • Service menu

  • Support for data infrastructure construction and business process reengineering

  • Analysis method

  • Interviews


Improving profits through inventory optimization using natural language processing techniques

The client was considering improving profits by increasing the accuracy of transaction history visibility. We applied natural language processing techniques to the matching of detailed product information with master data, which the client had not been able to link before. While increasing the accuracy of transaction results visibility, they succeeded in improving profits by identifying low-profit merchandise and circumstances giving rise to opportunity loss. This enabled a reallocation of investments and inventory distribution to emphasize high-profit products.

  • Value chain

  • Procurement /
    manufacturing

  • Data used

  • Transaction data /
    product data

  • Service menu

  • Data-driven consulting

  • Analysis method

  • Text mining
    (natural language processing)


Analyzing the causes of scandals to improve quality management operations at agency offices

The client had been promoting quality improvement activities at its sales agencies, and was considering business process improvement through data analysis. We performed analysis on agency operational quality data, and quantified the risk factors for occurrence of scandals. This enabled the formulation of a plan to improve the pattern of occurrence of such incidents, achieving more sophisticated agency quality management operations.

  • Value chain

  • Marketing / sales

  • Data used

  • Contract data

  • Service menu

  • Data analysis BPO

  • Analysis method

  • HyperCube analysis[a]


Optimization of hiring activities through construction of a hiring-related prediction model for sales staff

The client aimed to hire and retain as many excellent sales employees as possible. HyperCube analysis was used to construct a model that used hiring event participant data and other information to predict which employment candidates had the greatest probability of being hired and remaining stably in place over the long term. This made it possible to refine hiring activities.

  • Value chain

  • Personnel / organization

  • Data used

  • Policy data /
    event data

  • Service menu

  • Data-driven consulting

  • Analysis method

  • HyperCube analysis[a]


Recommendations for insurance product design descriptions leading to more efficient production and improved sales results

The client was aiming to reduce the operational cost of producing insurance product design descriptions, while increasing the associated policy sales closing rates. We used a nearest-neighbor calculation method that we developed, providing recommendations for insurance product design descriptions and constructing a searchable recommendation scheme.
We recommended insurance product design descriptions with the highest likelihood of policy contract sales for individual customers. Searches became more quickly executable, leading to major reductions in operational costs and also contributing to higher policy sales closing rates.

  • Value chain

  • Marketing / sales

  • Data used

  • Policy data

  • Service menu

  • AI technology adoption support

  • Analysis method

  • Nearest-neighbor calculation method(patented)


More efficient insurance quotes through construction of a tool to estimate policy cancellation refund amounts

The client was considering reducing the operational burden involved in providing individual existing policyholders with estimates of their policy cancellation refund amounts. We developed a tool by which policy data and calculated information enabled users to search current policyholder cash value and print it out in the form of a proposal, thus increasing operational efficiency.

  • Value chain

  • After-sale service

  • Data used

  • Policy data

  • Service menu

  • AI technology adoption support

  • Analysis method

  • -


AI voice analysis platform development

Recent years have seen increasing demand for analysis of voice data in such applications as AI-powered conversation evaluation and automated recording of conference proceedings. We therefore developed an AI voice analysis platform to handle tasks such as voice recognition and sentiment analysis. In addition to company-internal voice data analysis, we are moving forward with development of applications in which AI automatically evaluates the good and bad aspects of business talk using platform functionality.

  • Value chain

  • R&D

  • Data used

  • Voice data

  • Service menu

  • ABeam AI infrastructure provision

  • Analysis method

  • Deep learning


Ingraining governance activities by establishing operational systems for risk management data

The client was considering data governance activities focused on group risk management data. In order to ensure quality throughout the entire risk management data lifecycle, we set up operational systems including data dictionary compilation, data lineage documentation and data quality monitoring. This helped instill data governance activities companywide.

  • Value chain

  • Overall management

  • Data used

  • Risk management data

  • Service menu

  • Information governance build-up support

  • Analysis method

  • -


Individualizing debt collection methods to reduce costs and increase collection rates

The client was considering how to increase collection rates and reduce costs associated with billing reminders addressed to delinquent customers. We used AI to calculate response scoring, then individualized collection methods based on response scores, making it possible to reduce costs and boost collection rates.

  • Value chain

  • After-sale service

  • Data used

  • Contract data /
    customer data, etc.

  • Service menu

  • Data-driven consulting

  • Analysis method

  • HyperCube analysis[a] / deep learning / decision tree analysis, etc.


Improvement of sales staff performance through construction of an activity model

Given that the client was having difficulties in securing sales personnel, they were considering ways of improving the performance of individual sales staff members. We applied HyperCube analysis to daily report data (sales activity data), and constructed a model of activities of high-performance sales staff members. We used this model to enhance sales personnel guidance and education, making it possible to increase sales staff performance.

  • Value chain

  • Marketing / sales

  • Data used

  • Sales activity data

  • Service menu

  • Data analysis BPO

  • Analysis method

  • HyperCube analysis[a]


Using AI system implementation methodology to achieve short-term service release

The client sought know-how that would help them move forward with an AI systems development project aimed at achieving their goal of early release of AI services for retail use. We applied our AI systems implementation methodology to isolate and categorize critical items for consideration in project definition and design. Rapid implementation factors also included features such as the establishment of a process for accuracy evaluation, etc. By optimally organizing the results and project, they achieved short-term service release.

  • Value chain

  • Marketing / sales

  • Data used

  • -

  • Service menu

  • AI technology adoption support

  • Analysis method

  • -


Data analysis operational OJT support for data analysis personnel education

Aiming to establish a company-internal analysis organization, the client was considering how to train analytical team members. Since it was necessary to train members with little data analysis experience, we began the data analysis personnel training with analysis implementation led by our consultants. We then phased in client-led analysis gradually, rather than launching directly into having team members train and perform analysis. Rather than conveying knowledge in isolation, data analysis OJT was conducted along the lines of practical work, which succeeded in training personnel who could immediately apply their analytical know-how operationally.

  • Value chain

  • Personnel / organization

  • Data used

  • Customer data

  • Service menu

  • Strengthening AI capabilities

  • Analysis method

  • Random forest


Transportation / logistics

Realizing data-driven safety management operations through construction of an AI system for accident prevention

The client was considering how they could use accident-related data to prevent accidents from occurring. We applied HyperCube analysis and text mining to provide a system for easy-to-understand determinations of accident probability under specified conditions, and of factors that increase and decrease the likelihood of accidents. This made a data-based safety management system possible.

  • Value chain

  • Maintenance

  • Data used

  • Accident / safety-related data

  • Service menu

  • ABeam AI infrastructure provision

  • Analysis method

  • HyperCube analysis[a] /
    text mining(natural language processing)


Other

Risk visibility using a governance-risk-compliance (GRC) heat map

The client was considering more sophisticated GRC tools that could function cross-sectionally, traversing disparate company-internal groups. We applied our specialized knowledge and experience of statistical analysis and pattern mining to financial and non-financial data of each operational process, thereby quantifying risks associated with each operation to generate a heat map that brought cross-sectional visibility to business operations.

  • Value chain

  • Overall management

  • Data used

  • Financial data / non-financial data

  • Service menu

  • Data-driven consulting

  • Analysis method

  • Pattern mining, etc.


Improving product shipment plan accuracy with a demand prediction model

The client was considering how to move away from personnel-based product shipment planning reliant on the experience of each manager. By applying AI technology to predict model development, we were able to realize shipment planning with greater accuracy than before. Also, by making use of cloud-based AI visualization services, we were able to realize a system for complete and consistent execution of the PDCA cycle for data accumulation, prediction model construction, and prediction model improvement.

  • Value chain

  • Logistics

  • Data used

  • Product shipment data

  • Service menu

  • Support for data infrastructure construction and business process reengineering

  • Analysis method

  • Time series analysis


Work reduction, know-how accumulation and increased user satisfaction levels through construction of an AI chatbot

The client was considering how to reduce workload, accumulate know-how and elevate user satisfaction levels with respect to the handling of inquiries performed by personnel. We compiled issues to address in the field of travel expenses, and by automating inquiry FAQ responses using an AI chatbot, we made it possible to reduce workload, accumulate know-how and elevate user satisfaction levels.

  • Value chain

  • After-sale service

  • Data used

  • Inquiry history data

  • Service menu

  • AI technology adoption support

  • Analysis method

  • Text mining
    (natural language processing)


Reducing retirement rates by understanding trends among retirees

The client was considering utilizing data accumulated internally for the purpose of early detection of signs of impending retirement. We used HyperCube analysis to grasp behavioral characteristics associated with retirement, enabling an early approach to those who triggered concerns that they might retire, thus lowering the retirement rate.

  • Value chain

  • Personnel / organization

  • Data used

  • Personnel data

  • Service menu

  • Data-driven consulting

  • Analysis method

  • HyperCube analysis[a]


Building a sports performance analysis platform

In conjunction with Waseda University's Intelligent Performance Analysis Laboratory (IPAL), we initiated a joint research project on sports performance data analysis in order to promote the utilization of on-site sports data. Together with the start of the project, we constructed an analytics platform for the implementation of each form of analysis.

  • Value chain

  • R&D

  • Data used

  • Kinematic, psychological,
    neuroscientific data /
    performance data

  • Service menu

  • ABeam AI infrastructure provision

  • Analysis method

  • -


A service using Health AI solution to improve health and labor productivity

In order to improve employee health and productivity, we combined corporate operational information with employee health information to initiate a service that reports the results of statistical key performance indices (KPI) for health management together with correlation analysis results.

  • Value chain

  • Personnel / organization

  • Data used

  • Medical check-up data /
    performance data, etc.

  • Service menu

  • Data-driven consulting

  • Analysis method

  • HyperCube analysis[a]


Creating measures to improve employee productivity

The client was considering the creation of an environment and systems that would be easy for employees to work in. We used text data from employee satisfaction survey results, then analyzed work-style issues faced by employees with low satisfaction levels and those with low proportions of value-added activities. This led to the discovery of new issues that cause low productivity, and brought to light the need to inform employees about the personnel system. It thereby enabled the client to undertake effective remedial measures.

  • Value chain

  • Personnel / organization

  • Data used

  • Questionnaire data

  • Service menu

  • Data-driven consulting

  • Analysis method

  • Text mining (natural language processing) / cluster analysis


[a] Hyper Cube analysis: a method of multidimensional search analysis that uses exhaustive, comprehensive searches of the multidimensional data space to enable identification of the conditions giving rise to phenomena of interest.

Case Study

ABeam Consulting has delivered various kinds of data analytics projects using various kinds of data such as purchase history data, sensor data, text data, voice data, image data and so on, with cutting edge data analytics technologies. We can also help clients to utilize data in a full manner for all data related area such as people, organization, process, IT system based on our massive experiences.

page top