in2clouds is focused on helping companies use Predictive Analytics to improve their business performance. Founded by MicroStrategy alumni and launched in 2009, in2clouds is a small company that has been working in hi-tech, financial services and retail. Seeing analytics as “the next big thing” they want to reduce the friction for mainstream adoption and help companies achieve repeatable success – something that is hard for most companies trying to leverage analytics today.
In2clouds take a solution or decision-focused approach – focusing on defining the kinds of models, decisions and data needed in a particular vertical. They have developed their own technology to enable an effective predictive analytics program. A successful program in their mind requires a modeling and measurement framework, suitable data assets, technology, a focused business objective, business processes to deploy the models and organizational readiness for adoption. Like me they believe in beginning with the decision in mind – focusing on specific decisions before applying analytics. Each element has challenges and in2clouds tries to take a holistic approach to meet all these challenges. They try to partner with companies to move the program along across all the dimensions – optimizing for specific decision areas, offering SaaS/pay-as-you-go technology that has a wide range of integration options and allows the Force.com engine to be used for process management, and providing consulting help and continuous improvement of the models as needed.
In2clouds has three markets: end user companies, mostly business buyers, who purchase package solutions with pre-defined schemas, pre-built data connectors – a repeatable solution; they have OEM partners who embed in2clouds’ engine into their offerings; and corporate developers who are focused on an in-house application with an analytic component. End users are currently their main focus right now but the OEM partners are growing rapidly.
Their foundation layer leverages MapReduce and adds data extraction, data management, data enhancement and geocoding capabilities. They have developed some data mining, predictive analytics and adaptive learning routines including decision trees, neural networks, linear and logistic regression, self-organizing maps, time series and a Bayes classifier as well as some genetic algorithms for optimization and a champion/challenger framework. A layer of standard data providers (salesforce.com, US Census, Google, Twitter, other web services etc) complements this. A JSON/REST, SOAP API allows for real-time and batch access and they have some support for PMML in that they will output their models as PMML. They have packaged solutions for sales analytics, credit risk and value at risk.
From a UI perspective the solutions are obviously integrated with the business interface currently in use – the customer view is enhanced with analytics and decisioning for instance. The modeling environment has a web-based interface for each solution, primarily used to review models. For instance, a decision tree might be shown to see how wins and losses might be segmented – helping the business see the key selection criteria that drive results and preview results from a specific node in the decision tree. The ranking of, say, pipeline opportunities into deciles based on the score can also be investigated. Finally users get enhanced data in salesforce.com, adding in the probability of a sales opportunity closing for instance. For those developing their own solutions, OEMs, the various model review and data sampling tools are modular and web-based, making them easy to embed in the OEM’s own solution.