Fleet management is one of those areas where you could get a lot of benefits from data mining. In fact, one of the biggest draws of telematics in the field is the ability to gather and analyze huge data sets that could help companies fine-tune their operations and increase safety for their drivers and fleet. But how does data mining really help fleet management?
First, let us talk about what data mining is. Data mining is simply examining large data sets or databases to look for trends and new information. Data mining usually takes different approaches and perspective to be able to get a summary of new information.
You’ll have the best drivers and transport routes
Data mining is usually used to strengthen your bottom line by increasing revenues, cutting costs or a combination of both. But in fleet management, you can use data mining to retain your best drivers and manage your routes.
Predictive modeling can help you analyze data that you already have in your databases. This data could be used to learn more about your drivers and their behavior. For example, you might have a problem with driver attrition, you can check out the profiles of drivers who leave the company and see what they have in common. You might be surprised to find out that you could be telling your drivers to leave but you just do not know it.
For instance, a transportation and logistics company was examining its data to see how to improve driver performance and safety. By chance, they saw that drivers who get into an accident tend to leave the company in the near future. The company investigated why this is so and it was found that these drivers who got into an accident were sent an SMS message to their mobile phones with a stern warning that they had lost a certain number of points and had a set number of days to appeal. The company ascertained that this caused the drivers some degree of discouragement, and helping them decide to resign. The company changed the policy, stopped sending the canned SMS message and instead a driver manager met with the driver who had an accident and assured the drivers that they could very well make up for it.
On top of that, you could examine your data to improve your routes. A freight company can check its customers’ data to better understand delivery and pickup costs at each location. Knowing which customers and locations are more expensive, you can then reduce loading and unloading times at these locations to help improve your route.
Having better routes can also help you improve your customer service. Say you receive a call from a frequent customer and you would know which truck in your fleet is closest to the customer. This would allow you to dispatch a truck to your customer at the fastest time possible and this would involve the least cost for you to do so. So in effect, you are pleasing that customer at the least cost possible.
Better operational efficiency
It can also help you with your operational efficiency. Operational efficiency is simply being able to get the most out of your vehicles, and to do that you need data mining. Plus, it can also tell you when you need to put a truck under preventive maintenance.
The right re-selling price
Another area where data mining can help fleet management is in knowing how much to sell a truck for. So let us say that you want to sell some trucks in your field, how do you come up with the right selling price? This might sound simple, but it isn’t. Determining your truck’s residual value can be very difficult and a wrong calculation could bring some serious financial losses. With data mining, you can make use of the vehicle’s age, performance and condition in order to make more accurate calculations.
Operational Management and Management of Abnormal Conditions
These are just some of the ways data mining could help you manage your fleet. Transportation companies can actually use data mining in two very different situations. One is in operational management, or wanting to know things like if your drivers are following your stipulated routes or if they are regularly breaking traffic rules. Also, operational management might push you to look at which of your drivers are taking longer breaks than allowed, attendance and punctuality rates, and efficiency of delivery. Then you have the management of abnormal conditions, such as knowing things like whether your drivers go past the speed limit or goes away from the route.
Overall, most companies are looking to see patterns and information on normal fleet management.
Data and Its Analysis
On top of that, you actually use the data and the analysis in three different ways.
- First is data mining.
- Second is the traditional business intelligence where you get the data to create charts, reports, graphs and tables to keep track of performance indicators over a certain period of time.
- Then there is big data, where you not only take a look at your own data but data coming from different companies in your industry and data coming from different industries.
Data sources for data mining could come from different sources. Let us be clear that fleets by itself can generate a lot of data but these are data that can be pretty much very useless. In short, a majority of these data might not give you the information that you want. You certainly do not want to generate any more data, what fleet managers want is to get more insight from the data that they already have.
So the first step to look for data that you could use is to look at what you already have, or the data that you are already collecting. This might mean involving everyone who uses a computer to do their work. So the data you are getting are not only the ones that come from your fleet’s telematics system, although that would form a big chunk of the data you would work with. You will also look at the data you get from your planning people, your sales and marketing people, your work order management and other processes within the company. You could also get trip data coming from your truck management or maintenance systems. You could get data from your GPS systems, engine and speedometer.
So data sources is not the problem, chances are you already have a rich and deep set of data that you are getting and you could use these for data mining purposes.
And you are going to use all of these data in order to provide accurate predictive modeling. You could use predictive models to know which of your drivers are more likely to have an accident or when they are most likely to get into an accident. To know this for sure, you must collect all types of operational data, driving history, your drivers’ sleep patterns and other tools.
So what does that mean? Automate your data collection as much as possible. You can use great and inexpensive technology to do just that. For example, you can make use of handheld devices to make sure that not only are you collecting accurate delivery and transit data, but it is entered into your database automatically, without having anybody to enter it for you.
Analyzing and understanding your data
But even the best and most comprehensive data set would mean nothing if you cannot make sense of the data you have. Much less if you cannot access all the data you have to make actionable plans using it. You would need software, such as transportation management systems, mobile communications systems and fleet management systems.
You need to have a system that would gather all the data you have, put them all together and organize them in such a way that you could get insights and actionable items from them.
The thing is that you would need to collect all the data you get. This is because you simply cannot know what it is that you do not know. Meaning, in order to get new information from your data, you need to put your own assumptions on hold and just collect all the data you can. It might be overwhelming if you think about the volume of data you would need to store, but then again, there are data warehousing services that you can avail of.
This way, when you do run your analysis on the data that you have, you will be able to stumble across something new and something you did not know before, rather than get the wrong insights from incomplete data.
Some types of insights and information you could get from data mining, include:
- Causal effects, wherein B happens because A and data mining can give you a causal relationship between two things that you might not have expected.
- Clustering, wherein you can get groupings because of similarity.
- Trends, wherein you can detect similarities in a set of mined data. Like for example, driving later at night increases the chances of accidents.
- Sequences, wherein you could detect series of values or events.
So these are the basics you need to know about data mining and how fleet management can benefit from it.
Photos courtesy of TruckPR, gravitystorm and Toyota Material Handling EU.