Decision Science in Action

The objective of this article is to propose a design of a personalized recommendations application for restaurants owners which can be used by the customers of the given credit/debit card issuer. The developed recommendation application can be installed as mobile application in the mobile phone of the customers. The recommendation engine will use machine learning algorithms and predictive analytics techniques to derive information from  the credit transaction data, mobile location data, social media data (Facebook, Twitter), public data (Yelp.com/Bundle.com etc.), mobile browser data and data of the restaurants*. The recommendation engine will have two modules. First module will use the clustering algorithm to derive actionable information from credit card data and restaurants data. It will also use the data mining techniques to integrate the useful information available in the web space with restaurant data (explained later). This module will execute in server of the bank...

/* if the data set exists, then conditionally execute the step(s). %sysfunc checks the availability of the data in the library location. It means it checks historical sas data. It generates the system error if the not is data available at the defined library  */ %let dsname=&dataval.MKT_DATA;  /**** declaring data set name with library*****/ %macro warning1(name); %if %sysfunc(exist(&name)) %then %do; Proc printto log="&logfile.\WarningLog..txt" new;     /*** printing log in the text file at specifile location &logfile****/ run; %Put WARNING :- &name does not exists; Proc printto  ; Run; %end; %else %do; Proc printto log="&logfile.\WarningLog.txt" ; run; %put &name exists; Proc printto  ; Run; %end; %mend ; %warning1(&dsname); /**** Pros:  If any sas process is dependent on the existence of the data then we can use this macro***/ /**** Cons:  *Even if the SAS data set is a...

Modeling Practices of Loss Forecasting for Consumer Banking Portfolio Roll Rate Models The roll rate model is the most commonly used modeling practice for the loss forecasting in the consumer banking arena built at the portfolio level instead of at the individual account level. In this modeling practice, the whole portfolio is segmented by various delinquency buckets, e.g. Current, 30-DPD, 60-DPD, 90-DPD, 120-DPD, 150-DPD, and charge-off. The purpose is to evaluate the probability of an account in a specific delinquency bucket flowing into the next stage of delinquency status during the course of 1 month. The table below demonstrates the scheme of a basic roll rate model. In the table below, projected rolling rates are shown in the bottom two rows highlighted in red. The projected rate for each delinquency bucket is simply the moving average of previous 3 months. Due to its nature of simplicity, the roll rate model is able to fit into various business scenarios, e.g. d...

There are multiple OR packages to solve the optimization problems at academic as well as industry levels. One of the best OR packages that I am using is the SAS OR package. It has a capability to model highly complex problems and program it if required. Most of the optimization algorithms have been implemented on SAS.  In SAS we call Procedure to any function. There are many procedures which are SAS implementation of the OR algorithms. But there is a procedure called Proc Opt Model using which we can call different algorithms to solve any optimization problems. We can write  programs also inside it implement some constraints which can not be directly coded. I think this is very powerful tool for an operational research analyst. I use this procedure to model the business problems. I learnt it myself by going through user documentation. The SAS document explains clearly how can we use SAS OR packages with examples. In fact all the examples given in the HP Williams Model Buildin...

I recently worked with a cement manufacturer and a leading technology company to improve their supply chain network. Granularity of the two projects are very different. In case of cement manufacturer objective was to optimize the distribution and production planning. This was at a tactical level where as the objective of the work with technology company was at strategic level. Senior management was looking for some strategic level recommendations. But there are few things which were common such as issues with the data provided and structure of the results to be produced. Though structure of the results evolved gradually as we made progress on the project. I studied operation research during my post-graduation but was not using at my work place. These two project gave me opportunity to utilize my optimization skills to solve the business problems. While working in these projects, I learnt some new terminology as well as tactics to drive the optimization projects successfully. ...

There are several fraud claims in insurance industry. The manufacturing industry also suffers with similar fraud claims. The manufacturers sells the product in the market with warranty or guarantee of good quality. They also promise to replace or repair the product according to the agreed policy. Based on the extent of risk and product life pricing of the selling product is decided.  Manufacturers receive several warranty/guarantee claims in a year. Many of them are fraud claims. Identifying the authenticity of claims is rigorous and laborious process. But in the era of the machine learning and data analysis the cost and time of above process can be reduced.  I am discussing here a process that can be used to identify the veracity of the claim. This process framework is based on the machine learning algorithms. I believe that this framework has capability to reduce the time and cost required to investigate any claim manually.  The process flow of th...

In an e-retail setting when a consumer orders products, information are passed to brick and mortar warehouse for delivery of the ordered product.  The brick and mortar warehouse may be owned by same e retailers or may owned by a third party, serving the on-line demands as well as the in-store demands. As most of the in-store retailers have also started selling products on line the layout of the warehouse should be redesigned to reduce the operational cost. So that the replenishment time for the store and delivery time for the on line order can be minimized.  In a traditional warehouse design a separate layout for the delivery of the on line orders has not been studied. Most of the studies focus on the layout of the warehouse designed for the in-store retail type of sales. In this research we study the impact of the on line sales on the warehouse design. When an online orders are received at the warehouse through the website of the e retailer delivery process is initiated. T...