Companies perpetually try to reduce costs and optimize processes. At the same time, trying to optimize the conversion of customers through marketing efforts is also an optimization process. In as far as data is available and the aims well defined (i.e. within the context of data science and techniques ) a lot of business challenges can be casted as optimization problems.
Optimizations is a generic term for maximizing or minimizing some objective function. Examples are:
- maximizing portfolio return, the so-called Markovitz problem: allocating funds to stocks in order to minimize risk for a target rate of return (with known or computed variances and covariances)
- product mixing: determine how many products of each type to assemble from certain parts to maximize profits while not exceeding available parts inventory
- machine allocation: allocate production of a product to different machines, with different capacities, startup cost and operating cost, to meet production target at minimum cost
- transport optimization: determine how many products to ship from each factory to each warehouse, or from each factory to each warehouse and direct to each end-customer, to minimize shipping cost while meeting warehouse demands and not exceeding factory supplies
- media planning: decide how much advertising to purchase in different media to minimize total cost while achieving a target level of reach or frequency
- crop planning: given forecasted crop prices and growing conditions, determine how much of each crop to plant
- generator commitment: given forecasted demand by period and operating cost for each generator, determine which generators should be run in each time interval
All of these problems have a long history and solutions have been around in the shape of spreadsheet macro’s and whatnot. With big data and modern data science things have scaled in all directions delivering new and more precise solutions, effectively re-inventing business optimization.