Last week we visited both the Big Data Expo and CRMin1day. Then it was followed by a hackathon on Friday and Saturday, so it took a bit to digest all the information! I already posted some small pieces on LinkedIn, but I want to compare both events. My colleague Artem also chipped in with his experiences.
For people that don’t know both events:
- Big Data Expo. The Big Data Expo is held in the Jaarbeurs in Utrecht and entrance for visitors is free. It was about big data.
- CRMin1day. CRMin1day is a much more targeted event that was organized in Bussum. It was specifically targeted at people interested in CRM and user experience.
My and Artem’s experience of Big Data Expo was very different from CRMin1day, but for different reasons. I had a great time at CRMin1day, but at Big Data Expo I didn’t feel it. For Artem it was reverse: he liked Big Data Expo, but was a bit out of place at CRMin1day. The main reason was that everything was in Dutch at CRMin1day, and Artem has not yet reached his ceiling in Dutch.
However, the verdict on some aspects was unanimous.
- Big Data Expo had the coolest stuff for us personally. We love machine learning/A.I! (Pro tip: if your partner considers building an image recognition algorithm and program that solves sudoku’s automatically is more relaxing than solving the sudoku him/herself, it’s too late.)
- The variety of companies was nice at CRMin1day, it gave a nice overview of a field.
- At CRMin1day every stand had a lot of space and there was a nice central area with tables to stand. Big Data Expo was more crowded, which made it more noisy and less friendly for conversations.
- The food at CRMin1day and the drinks (some fruity-colourful-vegetable things) were constantly supplied in variety and there was excellent coffee. Great! Big Data Expo was free to enter, so understandably less awesome. I was told it’s obligatory to complain about coffee. Like complaining about the weather. So, in this spirit; the coffee was not great at Big Data Expo 😉
For me CRMin1day was excellent especially because our solution Journeylytics was very well received. Of course, we know it’s awesome, but still nice if people tell you they like it! Also, there are so many opportunities in the landscape of CRM’s and user experience, we already came up with several new ideas/products/services that would be a great fit. We had some people asking if they could get an demo, since they missed us at the day or we we’re all busy. Just send us an e-mail, and we’ll set it up.
Additionally, I got talking to the other sponsors of CRMin1day. Since we had ample time and space to chat, I met nice people and got to know a lot more. My compliments to the organization for this.
Artem had great discussions and ideas during the Big Data Expo. I’m sure he’ll write a blogpost about some awesome time series analysis later. There are many great ideas from econometrics that can be combined with ‘big data ideas’, and he’s going to do that exactly!
At both conferences we encountered good topics to start thinking about. I’ll just list them here:
- Lots about privacy. Next year there will be new EU-wide regulations about privacy: GDPR. People have to consent beforehand how their data is going to be used, and have the right to get their data deleted. In my LinkedIn post I made a remark on how we can use cryptography to still get value out of the data, while respecting the privacy. After some checking on the internet it seems the guidelines explicitly mention anonymization and pseudonomization. I still think there are great opportunities for more advanced techniques than plain anonymization/pseudonomizatio, for example, replace postal codes by random ones.
- Cleaning. Should we consider ‘spring cleaning’ of our data? Throw out the garbage, dust all our datasets and throw away anything not relevant. Or should we also keep all data? Despite that storage could be cheap, the mental effort could be expensive. For example, suppose you keep a rubbish data set of 2015 when your data collections was bad. There is no other data set, so why not? Well, from this point the reaction to any report can be: “So, how did this look in 2015?”, and you’re off to work with a horrible data set… If you simply accepted it was rubbish and only kept aggregates then we would have skipped pointless work. I understand that it is a choice to look at 2015, but unfortunately in my experience it’s always easier to argue for more information, then for less. More information is better right? Well. Apparently not 🙂
- Type checking of data pipelines. We had a cool idea that we can make data processing pipelines and type-check the entire chain! This can be done because our platform Rax is strongly typed. This means you can verify that (for tabular data) (a) the tables have all the correct columns (b) with the correct types (c) all operations correctly transform between types. Obviously, there are some harder things to check, like empty joins or business logic on ‘String’ columns. But it gives you lots of confidence that it’s going to work after changes.
That was it, our experiences of these two events. We are still going to do an evaluation of CRMin1day from a business perspective and do a follow-up, but that’s for later. In case you want to me tell me your experience or to give your opinion about this article, feel free to send me an e-mail or contact me on LinkedIn.