There are many approaches to calculating these kinds of data values. Some involve polling. Others involve data interpretation of a companies investor report. While these methods, and many more, all have their pluses and minuses, I'm going to outline today a fairly decent blended approaches to getting numbers that is both simple and desirable for accuracy with maximization of automation in a way that saves time without degrading the quality of the data.
A Blend of Technology and Curation:
Nothing can beat the human brain and its ability for decision making but three major faults exists in it in general. The first fault is human error bu that's a given in any situation. The second fault has to do with bias and the inability to accurate judge when certain external factors are present. However, in a situation dealing with harder mathematics that computes in terms of chronographic time and financial deltas, this isn't much to be worried about. The little bias that does exist is easily covered by revealing your testing methodology and leaving yourself open for peer review (to put it in scientific terms). The third fault is a problem here however. The problem is that of time. Time is of the utmost to many people, especially when dealing with a journalism site. It takes the human being precious time in order to make these decisions and that time can better be spent doing many other things, and the very investment of that time may make the task unwise or untenable for some reason or another because of its sheer magnitude.
This is where technology comes in to do some of the heavy lifting. In this particular case, web scrappers are best utilized. What is a web scrapper? Well a web scrapper is almost exactly what it sounds like. It is a piece of software (though also a process) that pulls data from one or more websites off the internet. It ultimately is utilized in all kinds of useful areas (science, advertising analytics, etc) but in this case it can best be utilized for the purposes of seeing how many people bought a game and are vocally unhappy about it. Going through common websites from forums to social media locations, one can pull data pertaining to a particular game from a particular time period and process it. Doing so first involves pulling the data and sorting it through mechanical means. If a hashtag was used, for example, to signal the preordering of a game then include it. If key words such as "bought" and "[GAME NAME]" are used in the same sentence, then use that information. When the mechanical processing is done is where the human being comes in.
At this stage, one of a few things can be done. Depending on how the data was categorized, it can be fairly easy to run through it and tally everyone up. If you can get a certain, near guaranteed, accuracy that the code that get you then you can simply rely on that and disclose the accuracy rate. You can also do a random poll yourself from a fraction of those individuals to come up with some manually calculated accuracy number given a sample size then tally up the total number of polled and captured unique examples of certification of dissatisfaction with purchase or purchase whole cloth, then you can get a very nice estimate that way and use it. Alternatively, you can use a service such as Mechanical Turk to get individuals that run through your data in a very cheap (talking literal pennies on the dollar) for repetitive manual tasks such as data sorting. This would require certain prepping though so that should be taken into account.
What do we get with all this?:
The result of all of this hard work is a calculation of how much money could have been saved collectively had individuals waited to purchase a certain game until it was at discount. You can sort by dissatisfied customers, customers period, or any other group of individuals. You can also pull from specific examples of peoples wording on the game and get exact URLs to manual look back on for screen grabbing purposes to insert in any given article. Perhaps, if you so wished, you could select some of the spicier comments on any given game by searching for comments with curse words in it. If you wanted sentiment but fewer curse words, you may would be best served looking for someone first using the curse word method and finding an individual/s that had other comments on the subject with less colorful language but maintaining the emotion. The possibilities truly are endless. Of course the numbers would be completely different would people, in such large numbers, wait to buy a game until discount. This too must be understood and stated when you give the final shocker number for the total amount of savings.
In conclusion:
You know why this is here
(for those of you that don't know why this is here: https://www.youtube.com/watch?v=3qSCQ1hOdzs 13:06-14:04)
Leave a comment: