Test store selection

Next, we present the basie ideas behind om methodology. Seetion 2.3 presents the optimization models used to form elusters and select a test store within each cluster and predict total season sales across all the stores of the chain from test store sales. Section 2.4 reports an apphcation of these ideas to a retailer speciahzing in women s fashion apparel, with over 1,000 stores nationwide. Compared to the current testing process in place at the retailer, our method would reduce the cost of stockouts and of merchandise left over at the end of the season by enough to increase profit by 100%. 

We describe a methodology for resolving the key decisions in designing a merchandise test how many and which specific stores in which to conduct the test and how to create a season forecast for the entire chain based on test store sales. We assume the retailer has (1) identified a set of products within a classification that they would like to test, (2) specified the time interval within which the products will be sold (the sales season), and (3) determined a test period during which the products will be offered for sale in selected stores to test their sales potential. 

We first describe the optimization model used to form clusters and select a test store within each cluster. This model is a specialized integer program known as the k-median problem, which we solve with the highly efficient algorithm given in Comuejols et al. (1977). 

Equation (2) enforces the condition that each store is assigned to a test store, (3) that we have exactlY k test stores, and (4) that stores are ohIy assigned to chosen test stores. Objective function (1) is structured to select k test stores that minimize the total overstock and understock costs if total test store sales are extrapolated to develop forecasts for stores assigned to it. To illustrate this point, let store i be represented bY test store j (i.e. Xjj = 1). Then... 

In developing merchandise plans, the retailer divides the year into two seasons fall/winter and spring/summer. The first week of the fall/winter season is the first full week of September, while the first week of the spring/summer season is the first full week in April. In their current methodology, they select 25 test stores whose total dollar sales are close to average store sales for the chain. Product tests are conducted at these stores over a 3-week period. To develop a season forecast and supply plan for the entire chain, total sales during the test period are divided by two factors estimated from past sales history. The first factor equals the proportion of season sales that are historically observed during these 3 weeks, while the second factor equals the proportion of total sales that are observed at these 25 stores. 

To provide a comparison, we also evaluated forecast errors and cost for the rules used by the plarmers at this retailer, the k median method based on store descriptors, alone and combined with sales mix differences, and two standard approaches to variable selection in linear regression, since the problem of choosing k test stores and a linear prediction function based on test sales at these stores can be viewed as choosing the best k out of n possible variables in a linear regression. Given actual sales Sp and test sales Sjp for i = 1,. . . n and p = 1,. . . m, we used the forward selection and backward elimination methods (Myers (1990)) to choose k out of the n test... 

Production Test. In a small-scale production test, tinplate containers with two commercially available enamels and two end-sealing compounds, which were selected from the results of this study, performed satisfactorily when packed with beef and ham. Beef was irradiated with 4.5-5.6 Mrad at 5, —30, and —90°C ham was irradiated with 3-4 Mrad and 6-7.5 Mrad at —30°C. For this production test, beef and ham were packaged in round tinplate containers and ham in Pullman tinplate containers, frozen and refrigerated products were shipped 1,200 miles by truck, and were gamma irradiated at various doses and temperatures. Irradiated products were shipped 1,200 miles in a non-refrigerated truck and stored at selected temperatures and humidities. The integrity of the cans was evaluated after storage for 10 days, 3 months, and 6 months. 

Flowever, omitting to take account of the temperature of products used or stored in sunlight when selecting materials can lead to unexpected rates of deterioration or failure. Conversely, the temperatures reached during weathering tests, both natural and accelerated, need to be considered when assessing the results. 

Many anthicid species are known to be canthariphilous [121]. After take up, males store the toxin in the accessory glands and transfer it as a kind of nuptial gift to the females. Many male anthicid species are characterized by ely tral exocrine glands which serve for excretion of cantharidin depending on the cantharidin titre. Similar to Pyrochroidae (see there) females test the cantharidin load of males before copulation and select those males which previously were able to incorporate this precious defensive compound from exogenous sources. 

The final portion of the GLP surveiUance inspection includes examination of records and reports as described under Parts 58.185, 58.190, and 58.195. To accomphsh this, FDA assesses the fachitys ability to store and retrieve study data, reports, specimens, and so on in a manner that maximizes their integrity and utihty. This must include an overview of how the firm maintains materials such as the raw data and the various specimens that are developed in the course of the study. The investigators must become famihar with the facility s archives regarding their location and accessibility. The individuals responsible for the archives must be identified and FDA must learn whether or not the archive is indexed and if the materials and records that have been transferred and stored elsewhere are appropriately identified. Furthermore, the procedures for adding or removing materials from the archives must be examined and individual test systems are selected randomly to determine that all raw data, specimens, and documents have been retained as required. 

Sodium naphthalide doping was performed using a selected polyazomethine with a THF slurry of sodium naphthalide and stirring for 24 hours. Thereafter the mixture was concentrated and the doped polymer stored at ambient temperature until needed. Testing data for -doped experimental agents not supplied by the author of the current invention. 

---