In order to realize optimized operation of systems, for example visual item ordering, scoring, etc., it is essential to design algorithms (data processing order protocol) based on both business objectives and data science.

About Algorithm

Data analysis refers to add consideration from various angles to a data set at a certain point, but in actual business data related to business is constantly being updated.

  • For example, when a customer visits to a store, then to provide optimized service and suggestions for the customer, calculate the type and rank of the customer based on the past purchase history data.
  • For example, a company deploying multiple stores, each store manager decides the daily stock purchase amount of each shop. Then develop a system to calculates automatically based on the demand and sharing prediction formula and presents the recommended purchase amount for store managers’ reference.
  • For example, in order to decide the optimum production volume / transport volume in the field of manufacturing and logistics. If some system calculates and completely decide every amount of production, we can not respond at unexpected exception. So rather good practice is to develop a system where the responsible person can judge referring recommended amount by the system, with a mechanism of simulation based on data statistics.

In these situations, it is not easy to order the software programming of the system that actually operates with the system development company, or to order from the development department inside the company, but to determine whether or not the success of the effect greatly depends on the calculation procedure firmly It depends on how it is done.

These calculation procedures are called “algorithms”. As a matter of course in designing algorithms, knowledge and ability of mathematics / information technology such as ability to design mathematical structure, or to understand data analysis for parameter determination are essential. At the same time, unless you understand the business structure of what to use for what purpose and what to maximize as an indicator of success, it tends to be mathematically clean but actually useless.

Service configuration example

We can choose the scope of coverage and role settings, taking into consideration what is necessary according to purpose and human resources held inside your company.

  • Case handling up to “conceptual design of algorithm”
    This case assumes situations where software engineering resource is available separately (in-house or outsourced) and can handle coding, and also mathematical concepts can be handled inside the company to a certain degree. Then, we can cover the roles of designing methods of assembling the procedure at conceptual levels, and also explaining its mathematical / spatial design philosophy, so that such design can be dropped into the code by the in-house or outsourced engineering resources.
  • Case handling up to “algorithm detailed procedure”
    This case assumes situations where in-house (or outsourced) software engineers can perform coding separately and detailed calculation procedure specification is necessary for coding process. Then in addition to the conceptual design of the algorithm, We prepare a calculation procedure documents which explains the data processing and calculation equations for each step. Typically such documents include several hundred lines of text and mathematical expression, with specification constants like coefficients to be implemented by the algorithm. Programming languages and environments to be used for coding can be chosen depending on already-implemented systems or in-house skill availability.
  • Case handling up to “algorithm calculation code”
    This case assumes situations where When in-house (or outsourced) software engineering resources are not available for coding the calculation process even with detailed specification documents. Then, with designated input format to be used as the input value of the calculation and the output data format to output the calculation result respectively, we prepares programming code to execute algorithm calculation process and provide the code as the deliverables.
  • Case to train internal staff and acquire experience through joint development
    Rather than leaving the whole design of the algorithm untouched, staff inside the company can participate and deepen the discussion while making it together. The co-operative design process goes through discussing various aspects like design choices and its pros and cons, resources necessary for implementation, necessary data acquisition, etc. In parallel, statistical methods and explanations of the concepts can be lectured. Such experience jointly participating in the algorithm development process is the best way to obtain the competitiveness to handle the cutting-edge data science capability for real business.
Service process
  • Status assessment, Method selection
    When you contact us, firstly we will ask you about the current situation. We will examine what is necessary now, what kind of method is best for the current purpose, etc.
  • Work design
    We explain expected deliverables, required period, cost, role sharing etc. and if you get internal approval etc, we will start the actual work.
    (NDA exchange if necessary at initial contract)
  • Algorithm construction
    We process conceptual design of the required algorithm.
    Mathematical structure design and parameter setting method etc. are consolidated through discussion, or by presenting a draft and being confirmed. We will consolidate the main framework with several meetings as necessary.
    The statistical and mathematical concepts used in the algorithm will be explained as necessary during such discussions.
    In some cases, additional data analysis is required to determine concept design and parameter setting method. In such cases we request additional data export and then we perform necessary analysis work. (typically for several weeks)
  • Detailed creation works
    In the cases where a calculation procedure documents are required as deliverables or when calculation codes are required, detailed calculation procedure for each step is actually created after the core procedure steps are settled.
    In such detailed creation works, tasks such as writing down mathematical concepts into the calculation procedure, inserting various exception processing, etc. should be covered. (typically for several weeks)
    If sample data is available, we can also use the sample data to verify the calculation procedure and its results.
  • Deliverables
    Finally, according to the setting of deliverables agreed, electronic files such as calculation procedure documents and codes will be provided as final deliverables.
Experience example
  • Manufacture and retail company
    Firstly created customer cluster map for marketing strategy formulation. As the next step we designed targeting product planning, promotion planning, CRM approach for each customer cluster. In order to realize these actions for customers, we needed to deliver different items for each customer, so to classify customers who visited website and real store, we created appropriate algorithm which reflected the early analytical result and understanding of customer characteristics.
  • Healthcare business
    Based on the information of a large number of users, analyzed factors that affected the risk of health and then created calculation formulas to index such factors. Then created visualization expressions to be displayed for individual users.
  • And many others
Contact

If you are interested in our services, please let us know from the contact form.

You can also find contact information at the company information page.