网络科技

    今日:340| 主题:245949
收藏本版
互联网、科技极客的综合动态。

[科技] Applying machine learning to the freight industry

[复制链接]
我受夠了等待 发表于 2016-9-30 12:16:07
208 13

立即注册CoLaBug.com会员,免费获得投稿人的专业资料,享用更多功能,玩转个人品牌!

您需要 登录 才可以下载或查看,没有帐号?立即注册

x
The current process of ocean freight forwarding is burdened with inefficiency, lack of upfront options and hidden charges. Kontainers is created based on the need for an easier freight forwarding experience.
  Kontainers is disrupting the freight industry with digitization and predictive analytics. We talked to Charles Lee, CTO and CPO at Kontainers, about their ideas and experiences in bringing paradigm-shifting forces to freight forwarding.
   
Applying machine learning to the freight industry-1 (technology,experience,objective,industry,learning)

  What's your role at Kontainers and what does that entail?

  (Charles: ) I’m the CTO and CPO of the company. I build and maintain the Kontainers platform that you see today. I set out the technology and product roadmap for the company and I’m responsible for delivery of that roadmap. I’m also responsible for acquiring customer feedback and make iteration improvement on the product.
  What does Kontainers do?

  Kontainers’ primary objective is to make freight shipping easier for business exporters and importers.
  What specific problems is Kontainers solving in freight shipping?

  At the moment, when you want to ship anything internationally, you’ll have to go through agents called freight forwarders. If you’re massive like Coca Cola, you will have your own logistics, but that’s only 30% of the shipping volume. 70% of the shipping volume has to go through a process that’s pretty much offline. Shipping a container can be around 20 phone calls and 60 emails. At Kontainers, we eliminate the countless phone calls and emails from the equation and provide instant quotes directly to you.
  20 phone calls and 60 emails. Why does shipping a container take so much communication?

  One critical piece of documentation in exports is the bill of lading. That alone constitutes to around 20 to 30 emails back and forth between the forwarder, the carrier and the exporter. Then there’s also communication between the exporter and the importer, because they need to both agree with the bill of lading. A lot of the time, a bill of lading doesn’t actually get finalized until the ship has arrived at the other end.
  You have loads of disparate systems and none of them talk to each other. It’s humans that do the talking between each system.
  Also there’s no central storage of documentation, and there’s a lot of documentation. You have loads of disparate systems and none of them talk to each other. It’s humans that do the talking between each system. When you need to find important documentation like a bill of lading, it’s pretty much just searching around the emails. It's a bottleneck and people have to spend a lot of time for a task that is crucial but very inefficient.
  So how are you guys solving these problems?

  So what we set out to do is to create an online platform that allows shippers to book the shipments electronically and digitally. At the moment our product allows you to make a booking with us in 5 minutes, and if it’s a repeated booking you can complete it in seconds. So this frees up a lot of time for our customers and they get to do what is more important to them, which is growing their own business.
  We have a system in place that reduces the whole conversation to two emails. Then the bill of lading is agreed upon by the exporter and the importer. It's that easy. No misunderstandings, no mistakes, no emails flooding your account. This is an example how one can use technology to streamline processes and break old habits.

Applying machine learning to the freight industry-2 (technology,experience,objective,industry,learning)

  What challenges have you encountered as the CTO at Kontainers?

  We’re still pretty much the only company that tries digitalization of an end to end shipment. So when you’re the first one, you often find that there’s a lot of stuff that needs improvement.
  From the technology standpoint, a big challenge is adapting the 40 year-old end to end shipment industry technology to suit today’s needs. The Shipping industry has been the most resilient industry to digital change. They don’t see a need for change. Freight forwarders don’t want to reveal information that customers can then use to bargain with them. That is a big reason why digitization is so slow. Carriers prefer to allocate money to keep the operations low-cost, rather than on digitization. It’s understandable because allocating resources towards digitization is more of a long-term investment.
  We’re still pretty much the only company that tries digitalization of an end to end shipment. So when you’re the first one, you often find that there’s a lot of stuff that needs improvement.
  One of the things we’re constantly working on tech wise is to offer more accurate shipping schedules to shipping line users.
  At anytime, we store around 25 million sailing schedules between 13 different shipping lines. A customer selects “from” and “to” locations on our site, and we give them a list of schedules arranged that have been stored in our system.
  One of the problems with this is that the technology the shipping lines use themselves are fairly backwards, and sometimes their schedules don’t get published into a distribution system until the boat has actually left. As a result, we cannot always list the most updated shipping schedules. Imagine if you want to fly from London to Paris. You looked at the schedule and it says that there’s a flight tomorrow. Actually, there was one today but the schedule just didn’t tell you. You can ring the carriers directly and find out about the updated schedule, but it doesn’t get into the distribution system on time to inform wherever you’re booking the flight.
  Kontainers is using predictive analytics to solve this problem...

  We’ve been collecting sailing schedules with these shipping lines for over a year now, it’s coming on two. We want to reverse the process with machine learning and predictive analytics. Instead of going to the carriers and passively wait for their schedule updates, we get the schedule, do some prediction and we might go back to them and say, “Hey guys, we think you are still sailing here, can you just do a confirmation?” We’re hoping to collect more schedules and improve the accuracy of predictions through monthly iterations. So possibly this can come to life in 2017.
  You can never avoid situations where a boat is broken down. But what you can do is to gather information on the engine of the boats, the fuel and ship types, all the information we have about a particular vendor, as well as weather data. Predictive analytics will give you some insights into whether or not a ship is gonna make it on time.
  Another thing that we’re working on using machine learning is customs clearing. Exporting goods out of the country is fairly simple comparing to importing. The customs requirements at the importing country are more detailed and are usually a hassle for our customers. The tax breaks and preferences at the importing country depend on the commodities’ country of origin and bilateral trade agreements. Customers usually need to provide documentations.
  We want to help customers save time and effort on filling out these customs forms. With machine learning, we’ll be able to automatically fill out the information needed for customs clearing by understanding previous import commodity data. When this feature is released, ideally the customer will put in country of origin and commodity, and based on these two pieces of information, we can auto suggest options and documentations for customers.
  What other challenges does Kontainers face in dealing with data?

  This is a very fragmented industry, which makes it difficult to understand market share. A lot of companies are onto acquiring customers because they can’t determine that they have 60% of the market by having a certain number of customers already. We need data analytics to help us understand our customer base better, to aggregate digital and offline effort in customer acquisition.
  It’s very important to get a holistic picture of where these efforts are. In a fragmented market, you’ll knock on a lot of doors, and if you can’t digest data from different sources, you’ll potentially miss out on a lot of potential clients.
  In a fragmented market, you’ll knock on a lot of doors, and if you can’t digest data from different sources, you’ll potentially miss out on a lot of potential clients.
  Right now we have a Hubspot subscription, but it’s expensive and doesn’t really allow you to input data. It’s not a data analytics platform, so it doesn’t allow you to push other data in from all the offline efforts. All the sort of knocking on doors and the data collected from there is dealt with in another CRM tool. So if anything, there are already two massive disparate sets of data source.This is where I see Traintracks can come in and help.
  What can we expect from Kontainers in the near future, besides the cool stuff you’re doing with machine learning?

  We launched imports into the UK in July 2016, hopefully we’ll allow imports into the US by the end of the year. Next year we’re looking into air freight. We want to turn Kontainers into a brand instead of a mere representation of physical containers.
  Bringing Data Culture to the freight industry…

  You know, in the end, our business is to help shipping companies. Data analytics is going to be a big feature for our customers to make sure that they get what they want, to go from A to B on time and with as small an amount of communication as possible. Because you think about their lives, shipping is only a very small part and it really shouldn’t take that much work.
   The Kontainers online platform instantly connects you to multiple carriers so that you can compare and select from the best rates and schedules in real-time. Check out their platform if you want to find cheaper and faster ways to do freight shipping. We’re looking forward to hearing more about their predictive analytics features in the next year!
  Co-contributors: Patrick Wenzek and Sarah Goh
友荐云推荐




上一篇:PSA: Today's update to the Oculus VR app on Samsung phones is causing heavy
下一篇:DAP计划即将实行,谷歌玩起了审核的套路
酷辣虫提示酷辣虫禁止发表任何与中华人民共和国法律有抵触的内容!所有内容由用户发布,并不代表酷辣虫的观点,酷辣虫无法对用户发布内容真实性提供任何的保证,请自行验证并承担风险与后果。如您有版权、违规等问题,请通过"联系我们"或"违规举报"告知我们处理。

ningningy 发表于 2016-9-30 12:38:04
求沙发
回复 支持 反对

使用道具 举报

ぁ詠遠∝僾妳 发表于 2016-9-30 12:38:38
你长的比假奶粉的毒性都大,我看了就头大。
回复 支持 反对

使用道具 举报

毒已深 发表于 2016-9-30 18:53:48
楼猪V5啊
回复 支持 反对

使用道具 举报

蔡文祥 发表于 2016-10-1 05:46:00
围观 围观 沙发在哪里!!!
回复 支持 反对

使用道具 举报

点赞 发表于 2016-10-11 12:45:45
不回帖的给脑残下了定义!
回复 支持 反对

使用道具 举报

Timothysr 发表于 2016-10-11 13:03:20
楼主呀,,,您太有才了。。。
回复 支持 反对

使用道具 举报

晓露 发表于 2016-11-3 17:25:06
我唯一的缺点就是钱多,而现在我几乎完美了。
回复 支持 反对

使用道具 举报

even2000 发表于 2016-11-15 13:39:11
东方不败外加灭绝师太啊!
回复 支持 反对

使用道具 举报

alilulu 发表于 2016-11-15 13:51:19
你会变成大海,然后被鲨鱼吃掉!
回复 支持 反对

使用道具 举报

*滑动验证:
您需要登录后才可以回帖 登录 | 立即注册

本版积分规则

我要投稿

推荐阅读

扫码访问 @iTTTTT瑞翔 的微博
回页顶回复上一篇下一篇回列表手机版
手机版/CoLaBug.com ( 粤ICP备05003221号 | 文网文[2010]257号 )|网站地图 酷辣虫

© 2001-2016 Comsenz Inc. Design: Dean. DiscuzFans.

返回顶部 返回列表