Surviving a Patrol Strike - Case Study Example

Your Your The Case of Centre Car-Park: Surviving a Patrol Strike Part The graph below represents the progression of parking revenues and fines in the City Centre Car-Park over the last 23 weeks, or almost six months. The pink line represents each week's revenue, and the yellow line indicates the amount of parking fines that are issued each week. Linear regression refers to the process of making a line fit as the explanation of a set of data at hand. A clear, linear relationship should appear, an example of which might be the way that a person's weight will increase (or decrease) based on caloric intake (Correlation 2003). However, linear regression requires that one of the two variables be fixed, which is possible in the current scenario. The yellow line (for fines) is fairly easy to deal with. If we eliminate the artificial blip of Christmas week and strike the clear outlier (Week 8) from the data, a trend appears. Smoothing would eliminate the faster spike right before Christmas, as well as the blip for the week when there were no fines, because of no enforcement. Over time, the car-park's traffic appears to be continuing to rise, even figuring in the seasonal effects of the Christmas shopping rush. This increase is leading to significantly higher revenues for parking fees in the weeks leading up to Christmas. However, both the higher changes in the weeks before Christmas and the lower results after Christmas do not suggest a business in trouble: instead they suggest a business that is in a highly competitive industry, City Centre Car-Park's market share appears to be growing, even if we smooth out the increases associated with the larger number of Christmastime shoppers. Part 2. As will happen occasionally in the labor market, the staff who are responsible for operating the car park have threatened to walk off the job after week 23. Because this staff is responsible for making sure the self-pay machines are locked and for patrolling the lot and issuing fines, this could have a significant effect on the car-park's revenues. While it is likely that people would still pay the machines, the question of enforcement and collection would be a sticky one. It is possible to use time-series analysis to figure out the approximate effect of such a walkout on the revenues of the car-park. The fundamental assumption of time series analysis is that the data being considered contain a systematic pattern, interrupted by error, or random noise, which can make the pattern difficult to find. Successful time-series analysis takes the random noise out of the situation as much as possible. The majority of time-series patterns consist of one of two basic types: trend and seasonality. Trend refers to a linear or nonlinear component that undergoes change over time and does not repeat within the time utilized by the model. Seasonality is a smaller version of trend, because it represents a cycle that repeats itself within the time utilized by the model. A set of data may contain both trend of seasonality. A common example would be retail sales, which may grow from year to year but may also be easily predicted to spike around the Christmas season within each year (Time Series Analysis 2002). While there is no established or proven way to find trend components within a set of time-series data, trends are fairly simple to identify, as long as they consistently move in one direction or another. When a set of data is considered to contain a significant amount of error, though, the first step is to try a process called "smoothing." This consists of averaging data within the set with the goal of canceling out the individual data that do not fit within the existing system. The most common way that smoothing occurs involves moving average smoothing. In this technique, each item of data in a series is replaced by the simple or weighted average of n surrounding pieces of data, where n is defined as the width of the "window" used for smoothing. Medians are often used instead of means here, for a couple of reasons. Medians eliminate the problem of the outliers in a data set. However, using medians eliminates the possibility of weighting in the average process. In cases where the error in measurement is unusually large, a process called negative exponentially weighted smoothing is used. In either case, though, the purpose of the technique is to take the data and create a smooth curve that is not affected by the outliers. Once this smoothing has occurred, it is often possible to approximate time-series data that moves in a consistent direction by using a linear function. If there is clear evidence of a nonlinear element to the data, the data must first be changed to remove the nonlinear factor from the situation. In most cases, a logarithmic, exponential, or polynomial function can be used to approximate the data involved. Seasonality involves a somewhat different analysis of the data. The formal definition of seasonality is "correlational dependency of order k between each i'th element of a series and the (i-k)th element. Autocorrelation, or a correlation between those two terms, is crucial to this process. The lag, or the amount of time that a particular seasonal cycle will last, is defined here as k (Time Series Analysis 2002). Correlograms can be used at this stage of data analysis to gauge the exact seasonal patterns involved. The correlogram will show the ACF, or autocorrelation function. The tricky part here is that the process of smoothing out one nonlinear element can have drastic effects on the rest of the data, if the wrong element is removed, potentially ruining the entire set of data collected. However, once seasonality has been determined, the process of differencing the series (taking each i'th element of the series and changing it to its difference from the (i-k)th element. There are two reasons to do this. First, this will reveal seasonal dependencies that have been hiding in the series. Because autocorrelations for consecutive lags are interdependent, each autocorrelation that is removed could result in the loss of another, even yet undiscovered, autocorrelation. The other reason to remove seasonality from the situation is the fact that the series needs to be stationary for ARIMA, or other techniques, to be used (Time Series Analysis 2002). And so a careful analysis of the data shows the detrimental effects that a strike would have for this small business. Since Christmas, collection of parking fines has headed down, with a drop of about 20% since the busy holiday season faded. While the higher traffic levels that one might expect right before and after the holidays appear to have dwindled slightly, as shown by the data in weeks 17-23, revenues would still be considerably affected by a strike by those employees who are in charge of making sure that the car-park's money is collected. Revenues from customers who paid for their parking spots has also dropped a little less than 25% than what they were making in the holiday season. A strike would take the amount of fines issued and push it towards zero just about immediately. If the public heard that the meter maids were going on strike, the number of honest people parking their cars and paying for the privilege will dwindle as well. However, an interesting trend involves the ways that parking revenue has stayed up above the '14,000 mark since Christmas ended. While some of this trend could well be people returning their gifts back (so they can spend more money in the same store!), increased awareness of the City Centre Car-Park, or other traffic functions could also be a factor. In any case, the revenues are not falling back down to the level in early June. One could easily assume a monthly loss of '2500 when the collection of fines stopped, because there would also be the risk of people parking and then trying to get away before having to pay for it. This additional risk might well cut the revenues in half, or even less. A liberal estimate would place revenues at '4,000 per month, simply based on those who are honest enough to go out of their way to pay the train faire without any overt supervision by the authorities. The linear regression with trend analysis for both revenues and fines appears below: By taking the median around the revenue windows, it becomes clear that there is a trend in this company that is headed upward, in terms of total cars parked in City Centre. While the Christmas retail season may have placed an artificial spike in the situation, the fact that the ending value for an acceptable minimum has gone up significantly since the time window began. Clearly, City Center Car-park needs to undertake an effort to avoid the strike by its employees. What has been, up to the present, a six-month improvement in revenues and fines, this improvement could quickly fade if the employees were not present. A lack of enforcement could lead to such concomitant concerns as people looting the pay machines to take what other people have put in there, more people simply parking and failing to pay, and a sharply reduced number of parking tickets issued, because of the lack of people to write those tickets (Cecil 2006). After a clear analysis of the wants and needs of the labor force, City Centre needs to take the time to make an informed decision about how serious it is about retaining its market share. Works Cited Bence J R 1995, "Analysis of short-time series: correcting for autocorrelation," Ecology March 1995. Accessed online 30 October 2006 at http://www.findarticles.com/p/articles/mi_m2120/is_n2_v76/ai_17041064 Cecil N 2006, "MPs call for an end to parking fines rip-off," Evening Standard, 20 June 2006. Accessed online 30 October 2006 at http://www.findarticles.com/p/articles/mi_qn4153/is_20060620/ai_n16491683 Correlation, and regression analysis for curve fitting 2003, Accessed 30 October 2006 online at http://helios.bto.ed.ac.uk/bto/statistics/tress11.html McDonald, R A, Behson S J, Seifert C F 2005, "Strategies for dealing with measurement error in multiple regression," Journal of the Academy of Business and Economics, March 2005. Accessed online 30 October 2006 at http://www.findarticles.com/p/articles/mi_m0OGT/is_3_5/ai_n16619674 Time Series Analysis 2002, Accessed 30 October 2006 online at http://www.statsoft.com/textbook/sttimser.html#identifying Read More