Depending on a firm’s business goal, one set of data can yield completely different analysis results. Guanghua professor Wang Hansheng sheds light on how a firm could useregressive analysis todig deep into data in order to maximize business values.

CORE APPEAL

Highlighting the key role of data analysis in generating values and even products, Wang, professor with the Department of Business Statistics and Econometrics under the Guanghua School of Management, urged a firm to first have a clear understanding of its core business appeal and its connection with the data available for analysis.

Wang recalled anInternet of Vehicles(IoV) firm tasked with analyzing data for a logistics company so as to determine “how good or bad their truck drivers are,” describing it as a negative example because “good” and “bad” are very ambiguous words.

“Data analysis is doomed to fail if a firm cannot say clearly what it wants. Only by making its goal as detailed and specific as possible can such analysis be accurate and helpful.”

After further communications, the IoV firm understood two biggest concerns of the logistics company : gas consumption and violation records. Data analysis finally revealed that some drivers have been stealing gas because, given the same type of trucks, same routes and similar driving durations, their gas consumption was obviously higher than others’.

According to Wang, it also helps to narrow down core appeals to only one or two, otherwise, it might end up getting less from analysis much like one failing to find a lover because he or she has too many criteria.

POSITIVE VARIABLES

Following a simple and clear goal, a firm should identify various factors that can positively affect this goal and apply them into data analysis so as ensure as better a result as possible.

“Some might say this process is redundant because the result will have been determined by the analysis model anyway. This is wrong,” Wang said. “In fact, models out there are quite similar, and those who have identified more positive variables will get more positive result.”

For example, sales records in a company’s online sales departments are usually dominated by the same several persons, and they might not want to share their success experiences with other colleagues in fear of being overtaken.

In order to spread their experiences to more employees, the firm’s boss could send some data analysis staff to be assistants at the side of these sales champions and thus give them opportunities to learn from them during day-to-day work so as to determine what affects their sales performances: the choice of sales targets, how they communicate or the way they advertise products. Eventually, these will be summarized into variables for data analysis and become an efficiency blueprint for the whole business.

According to Wang, firms should focus on variables they can alter (prices, ads, promotions, etc.) in order to maximize business values instead of wasting time on those they cannot.

In conclusion, data analysis requires a specific business scenario with clear core appeals and variables to generate maximum values. It also requires a firm to set aside pride or bias and give business the respect it is due.