Statistical population

You can perhaps think of a statistical population as a recordset (or a set of records). This set or group of records will be of similar items or events that are of interest to the data scientist for some experiment.

For a data developer, a population of data may be a recordset of all sales transactions for a month, and the interest might be reporting to the senior management of an organization which products are the fastest sellers and at which time of the year.

For a data scientist, a population may be a recordset of all emergency room admissions during a month, and the area of interest might be to determine the statistical demographics for emergency room use.

Another key point concerning statistical populations is that the recordset may be a group of (actually) existing objects or a hypothetical group of objects. Using the preceding example, you might draw a comparison of actual objects as those actual sales transactions recorded for the month while the hypothetical objects as sales transactions are expected, forecast, or presumed (based upon observations or experienced assumptions or other logic) to occur during a month.

Finally, through the use of statistical inference (explained later in this chapter), the data scientist can select a portion or subset of the recordset (or population) with the intention that it will represent the total population for a particular area of interest. This subset is known as a statistical sample.

If a sample of a population is chosen accurately, characteristics of the entire population (that the sample is drawn from) can be estimated from the corresponding characteristics of the sample.