This new model called: "Sherlock" is smart, speedy interactive. It really helps the media planner to find out the convenient plan. The time required for an evaluation is less than three seconds. For an optimisation it varies from some seconds to some minutes (usually 30 seconds). This range of response time makes it possible to use a fully interactive procedure to construct a plan. Each spot is selected very easily and quickly in a manual or automatic way. The plan under construction is evaluated at any time of the building process.
Typal analysis is a powerful tool for determining a structure within a large body of data . By studying and comparing the results of two series of experiments (random typology and typology of random numbers) we have been able to define two coefficients which measure informational content and the amount of structure for a given typology. The usefulness of these coefficients in solving the problems associated with typal analysis is demonstrated.
Manufacturers of electronic calculating equipment have recently produced high-output assemblies capable of reliably processing data at a very high speed. Despite the very wide choice of mechanographical equipment now available, the use of "electronic brains" is still comparatively restricted. This is because these machines have a very special language of their own, generally referred to as "machine-language". Any job first has to be arranged in a special form known as a: "program", consisting of a series of elementary operations or instructions which the machine is able to carry out automatically. This programme is prepared by a "programmer", who is conversant both with the machine language and the problem to be processed. His work takes up a certain amount of time between the designing and carrying-out stages of the process. It is also a possible source of error, and hence involves additional work in subsequently checking the calculations. Thus, it also inevitably leads to additional costs. Attempts have been made to overcome this disadvantage by making this "translation" stage automatic with the aid of special languages whereby the subtleties of machine language can be neglected. These languages are not entirely satisfactory for the interpretation of statistical data, both because their correct use by the operator requires considerable training, and as they still imply some knowledge of machine language. Furthermore, the translator does not directly carry out the statistician's instructions. These languages are very suitable for repetitive problems, which, however, only fairly seldom crop up in statistical data analysis, whore the process involved is usually one that is only carried out once. We have therefore developed an automatic translation system specially for statistical analysis, which does away with any intermediate steps between the statistician's and the machine's work. It consists of a special language which is used directly by the statistician and a "translator". We shall now discuss this language and its logic, followed by a brief despription of the translator's work.
Manufacturers of electronic calculating equipment have recently produced high-output assemblies capable of reliably processing data at a very high speed. Despite the very wide choice of mechanographical equipment now available, the use of "electronic brains" is still comparatively restricted. This is because these machines have a very special language of their own, generally referred to as "machine-language". Any job first has to be arranged in a special form known as a: "program", consisting of a series of elementary operations or instructions which the machine is able to carry out automatically. This programme is prepared by a "programmer", who is conversant both with the machine language and the problem to be processed. His work takes up a certain amount of time between the designing and carrying-out stages of the process. It is also a possible source of error, and hence involves additional work in subsequently checking the calculations. Thus, it also inevitably leads to additional costs. Attempts have been made to overcome this disadvantage by making this "translation" stage automatic with the aid of special languages whereby the subtleties of machine language can be neglected. These languages are not entirely satisfactory for the interpretation of statistical data, both because their correct use by the operator requires considerable training, and as they still imply some knowledge of machine language. Furthermore, the translator does not directly carry out the statistician's instructions. These languages are very suitable for repetitive problems, which, however, only fairly seldom crop up in statistical data analysis, whore the process involved is usually one that is only carried out once. We have therefore developed an automatic translation system specially for statistical analysis, which does away with any intermediate steps between the statistician's and the machine's work. It consists of a special language which is used directly by the statistician and a "translator". We shall now discuss this language and its logic, followed by a brief despription of the translator's work.