Concept: Predictive modeling is the most exciting aspect in the emerging and already highly sought after field of data analytics. It performs a learning process over the historical data set for a particular target to assemble a model which is enabled to forecast the future against the target. Our predictive models may be made up of a number of predictors, including variable factors that are likely to influence future behaviors or results. For example, given a click stream of an online customer and his/her demographic data, the model predicts whether the customer will churn out in the next six months. Another example, given the learning behaviour sequence of a student, a predictor calculates the probability that he/she may fail to pass the subject at the end of the semester. The resultant probability (risk score) obtained can be served as a predictor of academic failure.

IDA Technology: Our predictive models are embedded with the latest data mining, machine learning and statistics innovation to mine high-dimensional, heterogeneous and comprehensive relevant variables, with the delivery of discriminative predictors and indicators to best capture the trends or exceptions of data and behaviors. In government services such as taxation and social security, based on the payment behaviors and demographics, we can tell how long it may take debtor A to pay off the debt, while debtor B may take a different period of time. In banking business, by engaging data from social media like Twitter, call centres and banking information systems, we can predict who may likely move from a bank to another at what time for what reasons. In financial markets, by aligning data across capital markets, foreign exchange and petrol market, exceptional movements may be predicted to take place at what time in which market. In education, consolidating student engagements in class, library, online and campus, we can predict the academic performance dynamics of a student and recommend appropriate early intervention actions to convert low-performing students to the high-performing group.

Concept: It is hard to know enough about your customers and business from merely conducting empirical study on the raw data. To answer questions like "What customers want", "When they're going to buy/act", "How to optimize business resources", we can apply Customer Analytics, which deepens the understanding of consumer behaviors and marketing response, etc., by data-driven modelling to turn raw data into actionable customer/business insights. You will get benefits/savings from customized client service, effective market campaign, and optimized resources and so on.

IDA Technology:Different from traditional customer modelling and analysis which mostly rely on demographic and event data, our customer analytics models and algorithms deeply capture customer behaviors, intent, preference, life cycle, insights, as well as historical and future trends and development, etc. On top of business need, we can present customers in terms of life-span vs. real-time, specific vs. cross-business lines, local vs. global views, etc.

Concept: The in-depth analysis of human, organizational and system behaviours has been increasingly recognized as a crucial means for disclosing interior driving forces, causes and impact on businesses in handling many challenging issues such as behaviour modeling and analysis in virtual organizations, web community analysis, counter-terrorism and stopping crime.

IDA Technology: Behaviour analytics is to support explicit and quantitative behaviour understanding through a conversion from source data to behavioural data, and further conduct genuine analysis of individual or group behaviour characteristics, patterns, change, dynamics, utility and impact. It consists of key components including behaviour representation, behavioural data construction, behaviour impact and utility analysis, behaviour pattern analysis, behaviour change and dynamics, behaviour simulation, and behaviour presentation and visualization. Through behaviour modeling and analysis, we can:

Concept: Risk technology includes the systematic approaches to protect the things that we care about from potential loss. Risk is ubiquitous in all business domains, especially where decision are to be made. Risk identification, quantification and intervention are key aspects to understand and manage risk. Machine learning and data mining provide inherent power to understand risk through analyzing relevant data. For example, a fraud detection model can be built by iteratively and incrementally accumulating intelligence during the training process performed on the historical ground truth data set.

IDA Technology: We build risk models to detect and assess the risk, recommendation engine to support the intervention, dashboard to monitor the model performance, reporting portals to visualise the management of risk, knowledge base to assist the decision making. Skill sets like predictive modeling, behaviour analysis and big data visualisation contribute to risk identification, prevention, prediction and intervention from different perspectives which guarantee model robustness. Once identified, risky events or objects need to be removed/relieved by taking proper interventions. Action recommendation engine helps with this need. For example, different groups of customers respond to the debt collection arrangements in different ways. Our models quantify the effectiveness of arrangements, and suggest the next best action for a given arrangement and customer response. Our risk modeling approaches have been widely tested in government, financial organisations and educational sectors.