We have highly skilled professionals with knowledge in business domain and applied statistical techniques. Combining this talent pool with the IT, team building and project management skills, we have a successful formula for clients. Moreover, our association with renowned academicians & researchers ensures us the right guidance whenever required.

We have employed our skills & experience in various business domains such as Supply Chain, Financial Services (e.g., mortgage lending, collection & recovery), Online Advertising, Petrochemical Pricing, Commodities Trading, etc.

• Parametric tests such as z-test, t-test, ANOVA, etc.
• Non-parametric tests such as Chi-square, Kolmogorov-Smirnov, etc.
• Multivariate techniques such as Crosstab, Multiple Regression, Discriminant Analysis, CART, CHAID, etc.
• Forecasting using time-series (ARIMA, ARIMAX, ARCH/GARCH), Box-Jenkins

For some of our clients, we have used heuristic techniques such as Neural Networks, Bayesian Networks, etc.

We provide consulting using techniques from fields like Decision Science, Statistics, Predictive Analytics, Data Mining, etc. Being in the realm of advanced modeling requires us to perform complex calculations. Our calculations also require to be highly accurate and fast to provide accurate solutions at the right time for our clients to be rightly responsive. For which we use data mining suites such as SAS (currently the world wide leader in Business Intelligence Software) to help us gather intelligence and provide solutions to our clients.

Longitudinal Forecasting
When data is recorded in fixed intervals of time, time series methods are used for forecasting. The most simple of these is the exponential smoothening or moving average methods. An advanced technique like ARIMA gives a better accuracy of the forecast but involves a higher level of understanding and expertise. We have been using ARIMA for forecasting and our accuracies have always been higher than ninety percent.

Cross-Sectional Forecasting
The data used over here is independent of time and the dependent variable (or the forecasting variable) relies on other independent variables. Techniques like regression can be used which predict the dependent variable based on the values of the independent variable. Advanced techniques like logit or probit regression are also used based on the behavior of the data. For example, data in the retail space has high variance. Here a logit model would better fit the data than any other.

Knowledge Discovery from Databases
In the field of KDD, we have given solutions to clients from mortgage banks to debt collectors. We provide insights to our clients on their business with the help of Analytical Tools. For example, for one of our clients in the mortgage industry we were able to calculate Net Present Value for each of their products, identify cost centers in their marketing process, identify white spaces in the market, compare their market share with other competitors regionally and build decision trees for predicting defaulters.

Segmentation
Two types of algorithms are generally used for segmentation. One is the Supervised Learning and the other is Unsupervised Learning. In Supervised Learning, we have training data that is used to create a complex function. Whereas in Unsupervised Learning, a model is just fit to observation with no a priori output. Unsupervised Learning techniques such as Cluster Analysis are used in the initial stages of data modeling to identify clusters / segments. After the clusters are formed, the cluster characteristics are defined. Supervised Learning techniques are used to predict segmentation in a new dataset based on the segmentation in training dataset. For example, in a mortgage project we might use the previous years’ data as a training dataset for CART, which can be used to segment the present customers. We ensure that the segments that we recommend are SADAM (Substantial, Accessible, Differentiable, Actionable and Measurable).